Liar Loan
Well-known member
Kenkoko said:Liar Loan said:Kenkoko said:@Liar Loan,
If it's a randomized sub section / slice of a population, then yes, a 10% rate of infection would seem high currently.
But, you're not talking about randomized selection.
Would a 10% rate of infection seem high if we test the healthcare workers on the front lines? Not to me.
This is a specific company in a specific city with a owner funding a test because he just wants to know. Is it possible he already suspected he had infected employees? confirmation bias? the point is we don't know. This is also extremely small sample size given we have 3 million + in OC. Therefore this data isn't valuable.
You're right about false positive/negative being an ongoing issue. This is what happens when we allow the # of cases to go through the roof & not have the ability to test and re-test people who had exposure.
Bottom line:
There is some irrational exuberance on antibody tests. They are useful for tracking trends in the general population, but we are a long way from "herd immunity". Most experts say we wouldn't hit that until approx 60% of the population has recovered from the virus. We likely won't be out of the woods until we have a vaccine.
Nobody has suggested that we have herd immunity at this time. The point is that the fatality rate for COVID-19 determines the extent to which we can reopen society and get back to normal.
The antibody tests have a false positive rate that can be measured and accounted for. Let's stop wishing the past was different and focus on the tools we have at our disposal to gain better data.
415 actually is a valid sample size for a county of 3.2 million depending on the margin of error and confidence level you are going for. Just think about the political polls of 1,000-2,000 people that are used to gauge the feelings of 360 million Americans on a routine basis. Yes, this study is not a random sample, but it is a valid sample size.
This group of workers were not front line healthcare workers. They were IT workers that started working from home a week prior to the shelter in place order. You would expect a very low prevalance rate with this population, yet they tested at 10% positive. You can choose to disregard the results, or you can say 'Hmm...This warrants further investigation.'
That would be the scientific thing to do, but I gather the .
For a non-randomized and non-weighted test to be valuable data, it needs to be much larger than 415. No ethical healthcare professional would put much weight behind this.
This isn't even close to political sampling. You underplay the importance of randomization and properly weighted analysis. Should we call phone numbers from the Democratic data base to poll Trump's approval rating ? and present that raw data as valuable analysis? Surely you see how skewed it can be.
The scientific thing to do would have been to publish what kind of antibody test was used in this case. Do we even know how specific the tests are?
You're the one who said to extrapolate this data to the entire OC. Seems to me that science isn't really your agenda here.
Again, nobody suggested that healthcare professionals should put their weight behind this, or that you were unethical for not doing so.
What a good medical professional should do is pay attention to the evidence. That is why doctors have what is called a "practice". There may not be enough scientifically validated data for medical professionals to make good decisions during a novel virus-induced pandemic. They may be stuck relying on the best evidence available to them, which when you think it about it, is how we got 'shelter at home' in the first place.
So far, there have been 10 major antibody studies that I'm aware of, and all of them show higher infection rates than authorities originally believed, and if you extrapolate the results, none of them show a mortality rate above 0.60%, and most of them come in much lower than that.