Posted on 03/25/2020 8:43:21 AM PDT by Keith Gentile
Achieving Situational Awareness in the Mayhem of Coronavirus
A good place to start is where we should always start: Identifying and accepting what is most likely. And how do we discover what is most likely? As usual, it boils down to numbers.
Below are the numbers we need to be looking at. Which are we relatively certain of, what logical suppositions can we make, and what numbers may appear to be junk but could provide a good return if viewed from a new perspective?
The statistics:
1. Number of people who have been tested and confirmed to have Covid 19. This is about as sure as we can be provided the tests are accurate.
2. Number of dead: People today--as has been true for at least the last few weeks--who die from pneumonia-like symptoms in effected areas will have pretty routinely been checked for Covid 19. While test kits may have been lacking, these were amongst the types of situations they saved them for. Even if the person had contracted the disease and died outside the awareness of the medical system, they did die. Meaning they left behind a body that would have ended up with a coroner who would have had to establish a cause of death. And for the last several weeks, evidence leading to the obvious possibility of Covid 19 would presumably have been checked.
3. Time from symptoms to death: We now have enough information to determine that 2 1/2--5 weeks is a fair estimate. Here we have plenty of documented cases to go by.
4. Number of people showing up to be tested and then showing positive. Though we know the number, it has proven to tell us nothing about how many people are actually infected. How is this? The math just doesn't add up in any way we can bend our imaginations. Because 26 days ago (the average time between infection and death) we recorded exactly one new confirmed case when we can be reasonably certain total fatalities 26 days later was 111. Using the highest estimated mortality rate for Coronavirus (3% per 100) or lowest (.65% per 100), 111 fatalities today means we had between 3700 and 17,200 new cases per day 26 days ago. This at a time when only one had been reported! And guess what? That 3% highest level mortality rate was derived using that very same extraordinarily low confirmed number of infections data as representing the total number of actual infections.
But in order to obtain a "situational awareness" of this disease, we must be able to arrive at a viable number of total infections.
Understanding that using confirmed cases to represent actual total infections way underestimates the actual number of infections, obviously that 3% fatality rate is a vast overestimation as well. Therefore, it is far more likely that the lowest estimated fatality rate (.65) is closer to the mark. So if we trust that as the more accurate number--which indeed we must knowing what we do--that lower estimate resulting in 17,000 infections 26 days ago was more accurate as well.
From this we know the number of reported new cases data has absolutely zero to do with the actual number of infections. Does that mean its useless? Not at all. We can actual begin to make inferences using these confirmed cases--not at first to tell us how many infections we currently have--but at what rate the infections are increasing. After that, we can then use that rate to circuitously return to derive total infection.
In order to use confirmed cases of coronavirus as a tool to understand the past, current and future course of the disease, we must first remove variables.
Not knowing a virulent disease is on the loose means fewer people showing up to be checked, but rising with awareness. But now after two or three weeks, the seriousness and notoriety of coronavirus has widely been absorbed by the population, which is fast removing this variable for creating a false increase in hospital visits and consequent increase in detection making it appear the disease is spreading faster than it is.
But availability of testing may have decreased the rate of detection. Interestingly enough, the rate of increase in the total number of confirmed infections has remained fairly constant over the past few weeks, doubling every 2 1/2--3 1/2 days. If testing had been insufficient prior to this, it could only mean more cases should have been discovered meaning an even quicker doubling of infections was occurring But even after test kits arrived in sufficient numbers to test those asking to be tested, the rate of infection never seemed to change much. Nor did the rate change once the backlog of testing caught up. Only one of two conclusions can be drawn:
1. The real infection rate was higher than we thought when test kits were sparse and then dropped while the backlog of testing caught up producing a logical spike in detections and then rose again to cover up the return to normal detections after the backlog had been alleviated, creating the appearance of a flat line rise in infection rate as a result.
2. The other possibility, which seems more likely, is that hospitals were able to do a pretty good job of assessment through observation of symptoms who was most likely to be infected and consequently chose to administer one of the scarce tests appropriately on the truely afflicted. Inso doing, they were able to identify most of the infected walking through their doors. The real infection rate didn't rise and then fall and then rise again but remained at a consistent doubling every 2 1/2--3 1/2 days.
Confirmed detections of coronavirus only occur after people choose to go to the hospital to be checked. And here we have to make an assumption about people, when they will go to the hospital and under what conditions.
People who now go to be checked to see if they are infected with coronavirus might fall into three categories: those who consider themselves sufficiently ill to go to the hospital (with "sufficiently ill" covering an extremely broad range), those who feel ill at all, and those who are afraid they might be infected even if they have no symptoms. And once word is out that a potentially fatal disease is on the loose and everyone who fits into one of the three above categories is able to be tested, we might expect these percentages to remain relatively constant with only an increase in actual illness affecting them.
During this time, the same people who fit into the first two categories above (those feeling ill or sufficiently ill) are going into the hospital to be checked for the same reasons and under the same circumstances as they would have before. The increase in rate only being caused by the increase in illness (whether turning out to be coronavirus or not). Those afraid they may have contracted the illness but feel fine tending to remain a constant but might actual cover up the actual increase a bit. Therefore an increase in people arriving to be tested can reasonably be assumed to have been due to a rise in the infection rate of coronavirus, and the confirmed number of infections will correspondingly increase with the increase in numbers looking to be tested.
From all this we can conclude that the current rate of infection is probably close to rise in numbers presenting at the hospital to be tested minus a small amount for falling proportions of those "just checking to see" if they are infected.
And from this we can begin to extrapolate from those 17,000 cases we can be somewhat sure existed a minimum 26 days ago. Assuming an average doubling of infections every three days from what we just inferred over the past few weeks, 17,000 new daily cases would have transformed into an alarming total of somewhere around 8 million new infections just today with something around 13 million infections in total.
Estimates for total infections in "virgin soil" meaning amongst people with no history of a novel infection range from 1/3--100% of the population. If we average that to 2/3 of the population ultimately becoming infected, or 218 million Americans, we expect the midpoint when the infection rate will finally begin falling to occur around the time that number reaches 109 million, which gives us about 1 1/2 weeks to highest infection rate if the current rate holds. History shows us the falloff roughly to mirror the rise bringing the whole thing mostly to a close about five to six weeks from now.
If we look at the past as realistically as we possibly can right now and be honest with ourselves about the most reliable numbers we have and then have the fortitude to accept them and use them, things may look dismal . . . but maybe not so much. The most questionable number I've used in the calculations above has been the mortality rate. Knowing the actual mortality rate to be far lower than the rate of those showing up at the hospital and tested to be infected against those who eventually die of the disease, the Inferred Fatality Rate--derived for coronavirus only recently by one doctor in China--tries to factor this and many other variables into a more realistic prediction, in this case the .65% we used. But it is only when we assume this lower mortality rate is itself actually far, far higher than the actual mortality rate turns out to be, only now does it begin to explain why our confirmed infection rate from 26 days ago is so disconnected from our current fatalities. And this can mean only one thing: coronavirus may be far more widespread today than we thought, but it is also far less lethal.
Welcome to FR.
A reasonable, well thought out analysis of the situation.
Thanks for posting.
Not to mention that many places are releasing prisoners because of this. Seen two of them in the last three days just in my area.
Stay armed, safe and aware.
ping
Now THIS is how a new-comer should post. Great analysis - Thanks!
You mean we aren't all going to die from the sniffles?
Well done sir, well done!
Inferring from crappy data over three weeks back and then inferring AGAIN three weeks forward. Silly. We have better data. #new deaths and infer once.
It's a graph I ran across this morning. I found the perspective revealing.
If you have better data will you share it? Put it in a graph like this?
That would be great! ;)
What is “#new deaths” and how do I find it? Is it just a list of accumulated fatalities or is there any analysis?
Straight up raw data. Worldometers, the column in RED (hmmm, I wonder why?). There lies your best information about this disease.
Take the number of new dead and infer once. Much more accurate than starting with the worst data (new cases) and inferring back three weeks and then inferring a second time three weeks forward.
I’ve attempted repeatedly to enter Worldometer’s site, but Safari won’t let me. I wouldn’t suppose it has figures on numbers admitted to hospitals, would it? Can infer a lot of outcomes from that if only it were available. Knowing total infected is certainly good information, but total admissions would help in it’s own way as a tool.
I use Safari and have no problem. I feel so frustrated because I see graphs I wish I could share here that answer so many questions at a glance. I appreciate your efforts (if it was you that wrote the article). If you could have a little patience with me I could explain some things.
I would be happy to but if you are just going to call me names I wont waste my time.
It is strange. When I try to enter Worldometer I immediately jump to a Safari screen that (without recalling the exact wording) is a de facto stop sign.
And of course, please explain any insights you might have.
How do you reply so quickly? Before I even get back to the main page to see my own posting, I see yours reply above mine. I’m new to this site and I suppose I don’t understand exactly how it works.
And, yes, I did write the article.
Oh, and I won’t call you names.
OK. I went through this days ago but I can do it again. We have several groups of people providing us with data but which groups give us the best data? We know the asymptomatic give us the worst quality data and I think you listed the reasons. The mild may be a little better because people who are ill dont have to be chased down. The seriously ill is better because we have their bodies in the hospital. But the Dead is our best source of data. Sure there is such a thing as False Positive Dead but that doesnt mean what you might at first think.
So, we can agree (I hope) that the dead give us the best numbers. But there are lots of dead. Total dead, dead ratios, etc. but the best dead is #new deaths. Now we have to decide what do we do with the #? The way epidemiologists do this is quite different from how Finance guys might do so and for the following reason. Medically we want the information to provide us with just three things. We know from history that viruses are predictable. Especially new viruses that spread in a virgin population. In fact fewer things in life are more predictable.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.