Glock Forum - GlockTalk banner

1 - 7 of 7 Posts

·
Registered
Joined
·
7,547 Posts
Discussion Starter · #1 · (Edited)
https://sharylattkisson.com/2020/04...b4sIu2LXuRNTsYlXCUry2XzG7jFMdxeJUd84KlAfU6ojI

Knowing how many truly had coronavirus will allow scientists to calculate the first accurate death rate. Until now, the publicized mortality rates have been skewed higher than it really is because officials have only counted deaths among those who tested positive. A more accurate number will come from calculating deaths among everybody who had coronavirus, including those who were never diagnosed with it.

According to a preprint scientific paper on a part of the study conducted in Santa Clara, California, the population prevalence of COVID-19 in Santa Clara ranged from 2.49% to 4.16%.

That works out to a range of an estimated 48,000 to 81,000 people infected in Santa Clara County by early April. Those numbers are 50-85-fold more than the number of confirmed cases.

The scientists of the study say the population prevalence of Covid-19 (SARS-CoV-2) antibodies in Santa Clara County implies that the infection is “much more widespread than indicated by the number of confirmed cases.”

Coronavirus in Santa Clara County, California is “much more widespread than indicated by the number of confirmed cases.”

Study scientists


Small sample but I think this points to an overreaction thus far. JMHO.
 

·
Registered
Joined
·
1,803 Posts
There is a lot of iffy science in that study, IMO. Here is the preprint for those who are interested:

https://www.medrxiv.org/content/10.1101/2020.04.14.20062463v1.full.pdf

First, note that the study is as yet not peer-reviewed. If I were peer-reviewing it I would make the following observations.

First, the estimates of sero-positivity that these investigators are hanging their hats on is based on positive test results from exactly 50 persons who tested positive out of a final test group of 3330.
Extrapolating observed results from 50 people to a population of 1,928,000 people is fraught with potential error under the best of circumstances. And the circumstances here were far from ideal.

First off, their raw rate of sero-positive tests was 1.5%, not 2.49 or 4.16%. They arrive at those much larger estimates through some mathematical juggling that includes a multiplier that weights the observed rate by an assumed percentage of false negative test results, and then re-weighting the inflated results by the test subjects race, ethnicity, zip code, and sex to "adjust for" what they perceived to be an over-representation of white women and an under-representation of Hispanics, Asians, and men in their test population.

Extrapolating observed results in such a small test population to a population more than 500 times greater in size might have validity if the test population is truly representative. This test population was likely anything but representative of the greater population. The test population was not randomly-selected. It was self-selected. The test subjects were largely recruited through Facebook ads. Consider two individuals. The first has either had symptoms that might have been due to covid-19 infection, lives in a neighborhood that experienced a lot of cases, or had close contacts who had symptoms or tested positive. The second individual has had no symptoms, doesn't know anyone in their neighborhood who has gotten sick, and has had no close contacts with confirmed cases. Which individual is more likely to volunteer to go get tested for covid-19 antibodies?

In other words, there is very good reason to believe that there is a large element of selection bias in this small test group strongly favoring individuals who are more likely to have been exposed to the virus. If so, the test group is not representative and the results cannot be extrapolated to a much larger population.

Lastly, there are big questions regarding the validity of the testing methodology. The authors mention that the lateral flow assay they used comes from Premier Biotech of Minneapolis, MN. What they fail to mention is that the test was actually designed and is manufactured in China by Hangzhou Biotest Biotech in China. They provided the kits to Premier Biotech which then only markets them. The data for test kit sensitivity and specificity that the investigators are relying on come almost exclusively from what that Chinese company has claimed. And what is more, China's National Medical Product Administration (NMPA) which is roughly equivalent to the FDA in China has compiled a list of covid-19 test kits that they approve for export in order to "strictly control quality, maintain export order, and crack down on counterfeit and shoddy products" and this test kit is not on the list. The export of this test kit was banned by the NMPA on March 31. The kits that were imported into the US came in after the FDA relaxed its guidelines for import of test kits on March 16 and before the NMPA banned their export two weeks later.

That is not to say that the test kit is automatically no good, but China is known to have sold millions of dollars worth of covid-19 test kits to Spain, the Czech Republic, and the UK that proved to be completely worthless, so there is reason to be skeptical regarding the quality of test materials coming out of China and the validity of their claims for sensitivity and specificity.

The authors of this paper did do some testing of sensitivity and specificity of this test but the test groups were so small as to be laughably lame. To test specificity, which is the likelihood that a positive test truly indicates antibodies to covid-19 and is not a false positive, they tested serum from 30 patients who had undergone hip replacement in the pre-covid era. Fine, but with a raw positive test rate of 1.5% even if every positive test was a false positive we would expect to turn up exactly .45 people who had false positive test results in a group of 30 individuals. So the authors are basing their assumed specificity for this test on what the Chinese have told them.

And specificity can make an enormous difference when it comes to estimating the number of true positives in a huge population from a very small 50 person test group. False positive test results could be the result of circulating antibodies to protein antigens from viruses similar to covid-19 but not covid-19. If even a small percentage of their 50 positive test results were false positives, it would have an enormous impact on the percentage of sero-postive individuals in the greater population. The authors admit that if the true specificity of the test drops to below 97.9% their estimate of covid-19 sero-prevalence in Santa Clara County would drop below 1%. A single false positive among their 50 positive test results would drop the specificity to 98%. Two would drop the specificity to 96%.
 

·
Registered
Joined
·
162 Posts
I am 99% sure my wife and I had it starting on jan 12th here in Ma. We missed 11 days of work but did not seek medical help. We had 100% of the symptoms.
This virus was going around long before they started testing for it.
 

·
Registered
Joined
·
2,460 Posts
From reading any story about covid realize they will likely be discounted many reasons.

The actual benefit of a story or study like this is it could foster interest in more follow-on studies.

Could be something to or maybe not.

I take anything in the media with a huge amount of skepticism.
 

·
Registered
Joined
·
2,064 Posts
There is a lot of iffy science in that study, IMO. Here is the preprint for those who are interested:

https://www.medrxiv.org/content/10.1101/2020.04.14.20062463v1.full.pdf

First, note that the study is as yet not peer-reviewed. If I were peer-reviewing it I would make the following observations.

First, the estimates of sero-positivity that these investigators are hanging their hats on is based on positive test results from exactly 50 persons who tested positive out of a final test group of 3330.
Extrapolating observed results from 50 people to a population of 1,928,000 people is fraught with potential error under the best of circumstances. And the circumstances here were far from ideal.

First off, their raw rate of sero-positive tests was 1.5%, not 2.49 or 4.16%. They arrive at those much larger estimates through some mathematical juggling that includes a multiplier that weights the observed rate by an assumed percentage of false negative test results, and then re-weighting the inflated results by the test subjects race, ethnicity, zip code, and sex to "adjust for" what they perceived to be an over-representation of white women and an under-representation of Hispanics, Asians, and men in their test population.

Extrapolating observed results in such a small test population to a population more than 500 times greater in size might have validity if the test population is truly representative. This test population was likely anything but representative of the greater population. The test population was not randomly-selected. It was self-selected. The test subjects were largely recruited through Facebook ads. Consider two individuals. The first has either had symptoms that might have been due to covid-19 infection, lives in a neighborhood that experienced a lot of cases, or had close contacts who had symptoms or tested positive. The second individual has had no symptoms, doesn't know anyone in their neighborhood who has gotten sick, and has had no close contacts with confirmed cases. Which individual is more likely to volunteer to go get tested for covid-19 antibodies?

In other words, there is very good reason to believe that there is a large element of selection bias in this small test group strongly favoring individuals who are more likely to have been exposed to the virus. If so, the test group is not representative and the results cannot be extrapolated to a much larger population.

Lastly, there are big questions regarding the validity of the testing methodology. The authors mention that the lateral flow assay they used comes from Premier Biotech of Minneapolis, MN. What they fail to mention is that the test was actually designed and is manufactured in China by Hangzhou Biotest Biotech in China. They provided the kits to Premier Biotech which then only markets them. The data for test kit sensitivity and specificity that the investigators are relying on come almost exclusively from what that Chinese company has claimed. And what is more, China's National Medical Product Administration (NMPA) which is roughly equivalent to the FDA in China has compiled a list of covid-19 test kits that they approve for export in order to "strictly control quality, maintain export order, and crack down on counterfeit and shoddy products" and this test kit is not on the list. The export of this test kit was banned by the NMPA on March 31. The kits that were imported into the US came in after the FDA relaxed its guidelines for import of test kits on March 16 and before the NMPA banned their export two weeks later.

That is not to say that the test kit is automatically no good, but China is known to have sold millions of dollars worth of covid-19 test kits to Spain, the Czech Republic, and the UK that proved to be completely worthless, so there is reason to be skeptical regarding the quality of test materials coming out of China and the validity of their claims for sensitivity and specificity.

The authors of this paper did do some testing of sensitivity and specificity of this test but the test groups were so small as to be laughably lame. To test specificity, which is the likelihood that a positive test truly indicates antibodies to covid-19 and is not a false positive, they tested serum from 30 patients who had undergone hip replacement in the pre-covid era. Fine, but with a raw positive test rate of 1.5% even if every positive test was a false positive we would expect to turn up exactly .45 people who had false positive test results in a group of 30 individuals. So the authors are basing their assumed specificity for this test on what the Chinese have told them.

And specificity can make an enormous difference when it comes to estimating the number of true positives in a huge population from a very small 50 person test group. False positive test results could be the result of circulating antibodies to protein antigens from viruses similar to covid-19 but not covid-19. If even a small percentage of their 50 positive test results were false positives, it would have an enormous impact on the percentage of sero-postive individuals in the greater population. The authors admit that if the true specificity of the test drops to below 97.9% their estimate of covid-19 sero-prevalence in Santa Clara County would drop below 1%. A single false positive among their 50 positive test results would drop the specificity to 98%. Two would drop the specificity to 96%.
Please extrapolate and provide more detail
 

·
Registered
Joined
·
10,376 Posts
It appears the numbers presented on news outlets is cumulative. If they would report active cases instead of every case that’s ever been, it would be more telling of what’s happening now.
 
1 - 7 of 7 Posts
Top