I’m not sure ‘either lived or’ belongs there since we’re talking about excess deaths. But excess deaths would of course include those who were not infected with SARS-CoV-2 but died because they did not seek the medical attention they should have but didn’t because they were fatally reluctant to do so because of potential exposure to COVID – strokes, heart attacks, other pneumonias, even appendicitis, etc. (Three ‘becauses’ in one sentence is excessive. 
)
2 Likes
I don’t have all your answers at this point, (the clinical issue really does have good criteria but I don’t have time for it now)…it is irrelevant at present because testing is now very accurate and cheap. That clarifies things a lot. Thanks.
1 Like
Do you know how that is verified? For instance, this article says that even if a test has high specificity and sensitivity, it will still have a very high false positive rate if the infection is only a few percentages of the population.
In the US I think we have about 7 million cases, which is about 3% of the population. So if our tests are equivalent to the Cellex test, we could still be seeing about a 50% false positive rate.
1 Like
I am not sure of the question. The letter to the editor is talking of screening the general, asymptomatic population with IgG antibodies and discussing cutoffs to determine positivity. We were, I thought, discussing PCR for those with a very high pretest probability. IgG testing in the general population is not being used to diagnose the disease on a widespread basis. The article addresses a theoretical scenario. Up to Date and, I think, NIH and WHO and CDC would have more on that. Applied epidemiology illustrates where statistics and biology meet. I would have to dig those up, but that is the case with all testing…the lower the pretest probability, the more false positives (like with HIV). Or am I misunderstanding totally?) Sorry. Thanks.
1 Like
Whether PCR or some other test, I think the author’s numbers are correct. I read the wikipedia page on sensitivity and specificity.
- Sensitivity measures the proportion of positives that are correctly identified (e.g., the percentage of sick people who are correctly identified as having some illness).
- Specificity measures the proportion of negatives that are correctly identified (e.g., the percentage of healthy people who are correctly identified as not having some illness).
Sensitivity is how many with the disease are correctly classified. Specificity is how many without the disease are correctly classified.
So, in a situation with 100 people and 3 have covid-19, if our test identifies those 3, along with 3 others, totally 6 identified as having covid-19, then the sensitivity is 3/3=100% and specificity is 93/97=95%. Sounds like a really great test, but half of those identified with covid-19 are false positives.
Hypothetically, if we are using PCR to diagnose whether patients have covid-19, and it is only in 3% of the population, and our tests have equivalent sensitivity (100%) and specificity (95%) as in my scenario, half of the positive diagnoses may be false. Despite the tests having really great false positive (sensitivity) and false negative (specificity) rates.
2 Likes
You are right, but we are not testing random folks. Leaving for a walk and evening with kids (strep is going through the family…interesting case in point. My son was diagnosed positive by a rapid strep with only 85% sensitivity (I forgot the specificity), but if we get the same symptoms, it is considered accurate enough that the rest of our family who live in the same house can have treatment without testing. We all have gotten the symptoms to a lesser degree. The positive predictive value of the history and exam is enough to treat without the test. However, it would be a terrible test to use on asymptomatic individuals, as the prevalence is low in the general population. The relative sensitivity and specificity are worse than the Covid IgG. ). Maybe we can get into this more (or ask others to comment).
HIV is another great case in point. We don’t discount test validity or stats in appropriate circumstances of high pretest probability because of some false positives in the general population. Have a good evening. Maybe others like @glipsnort can discuss better. Thanks for iron sharpening iron.
This is an example where pre test probability and predictive value come into play. Good discussion. Thanks
3 Likes
You can use a medical test calculator: Diagnostic Test Calculator
As you both are discussing, it depends highly on the pretest probability. If you test people randomly then maybe you could use a table like this, otherwise the table is a little misleading as you could have a test with only 80% specificity, but the false-positive rate could be only 12% if your pretest probability is 60%. This doesn’t have to mean that 60% of the population has the virus, but as @Randy mentioned, if someone has clinical characteristics of COVID-19, that would raise the pretest probability. You also can raise the pretest probability by giving someone multiple tests. I made a boring lecture on the topic a few months ago:
5 Likes
There is quite a bit of essentially random testing going on, for example at workplaces and colleges that are testing everyone regularly. At the Broad, as of a week ago we’d run 400,000 tests from colleges, with a positivity rate of < 0.1% – so the specificity of our testing is clearly much better than 97%. To achieve that I believe we’re testing every positive sample twice (or maybe every sample – I dunno). New York City is reporting a positivity rate of 0.8%.
7 Likes
If we have had 7 million confirmed cases, we likely have more than 80 million total cases.
Covid patients take more resources. Our local hospital said it takes 10 times as much resources as a non-Covid patient. Another local hospital mentioned the payments and the fact that even with the extra payments, they’re losing money on Covid patients. They take so much more resources and they stay much longer, often several weeks.
Suggesting that hospitals are inflating numbers is to suggest that every hospital across the country is committing fraud. Surely you don’t think that’s happening, right?
7 Likes
The refrigerated morgue trucks probably don’t cost that much to rent. (I hope that was detectable irony.) I saw somewhere yesterday or today that a COVID patient finally got discharged from the hospital after 60 days.
3 Likes
The “true toll” isn’t something that can be ascertained with the precision and accuracy you want. I understand the desire to be able to pinpoint what that number actually is, but it is just plain not possible. The @Chris_Falter suggestion of confidence intervals is interesting. What we can say with certainty is that the “true toll” is much closer to the numbers reported by entities like Johns Hopkins (here) and the COVID Tracking Project (here) than the QAnon figures.
It is demoralizing and disheartening that our Christian brothers and sisters need much more help discerning truth from lies than the rest of the population, but that is absolutely where we find ourselves. Thanks for the efforts you have already made.
8 Likes
And MacArthur does nothing to mend his ways:
https://www.wthrockmorton.com/2020/09/06/john-macarthur-doubles-down-on-no-pandemic-claim/
5 Likes
As I said the other day:
It’s not bad to be wrong or admit your ignorance. What is bad is to confidently proclaim you are right despite correction time and time again on a topic by others who know more than you.
7 Likes
I suspect the structure of American Evangelical churches helps develop the sort of attitude displayed here. When there’s basically one person telling the congregation the Truth and he’s not answerable to anyone – it’s pretty easy to develop this kind of arrogance.
8 Likes
Yes, church government* and accountability is an issue. But maybe part of it is that it’s also an American issue, where everyone has all these unassailable rights, or so they mistakenly extrapolate, and they are ‘free’ to do what to do whatever they want, whenever and wherever. We’ve had discussions here about Law and law, love and rules, ‘love rules’ vs. laws of love. American Christianity has lost a lot with respect to the last in that series, but it is a perennial issue with being human.
*It is analogous to what is the best kind of civil government – there are none that are perfect. There are strict hierarchies, e.g. Roman Catholic, and there are also representative governments – Presbyterian (an Independent Presbyterian Church is an oxymoron ) and conference type structures, and the ubiquitous independent church structure with elder boards.
“Anti-intellectualism has been a constant thread winding its way through our political and cultural life, nurtured by the false notion that democracy means that ‘my ignorance is just as good as your knowledge.’ ”
-Isaac Asimov
7 Likes
Too true and well stated. (I love The End of Eternity. 
)
1 Like
It was not supposed to be! 
These dataset contain protected information (HIPAA rules apply) and are unlikely to be publicly released. If they were released, very few people are familiar with competing risks analysis to use them correctly. Even so, the competing risks would be coded per the current standards, and so the estimates ought to be the same as what the biostatisticians are saying.
And that’s the basic standard medical professionals used for everything. If you want to reassess that standard, you need to rework it for all possible causes of death, then make an argument for the replacement of the current standard. Even if this were done, my intuition is that is it would not make a big change from current estimates for death rates of COVID.
7 Likes
Do you know any biostatisticians? 
2 Likes