Thursday, July 23, 2020

Meaningfully discussion what is safe or not safe requires correct information

Let me start by saying this by saying I am not an expert in epidemic modeling. I do have expertise in statistics, I do have graduate classwork and some professional experience with epidemic data analysis and modeling. And in spite of that I am NOT an expert. Which is why when I want to know what is most likely to happen next, I read and listen to the words of experts. What I can do is explain how a lot of the reasonable sounding, statistical sounding arguments floating around out there are either wrong, make unjustified assertions, or both. To illustrate I will use some words used to argue for five days per week in person school for everyone who wants it in the school district where I teach. My district is the Dublin City Schools, in Franklin County, Ohio. 

We are not in level 4, we are in level 3 and they removed the star last week that warned we might go up to 4. Franklin County is increasing in cases, but has had a downward trend in deaths for 8 weeks, and hospitalizations for 7 weeks. 

OK, let's address the "big claim" here, that cases are up but deaths are down. Here are the graphs of the seven day rolling averages for cases and deaths. 
So on it's face this looks kind of like a true claim. It is true that we have learned a lot about treating this disease so it has become somewhat less deadly. The problem is the same one we are having right now with people arguing that the death rate is dropping dramatically nationwide. When they say "death rate" they mean case fatality rate. And the reason the case fatality rate is dropping dramatically right now is not that fewer people are dying but that the number of cases is rising quickly. On the other average deaths are taking almost 2-4 weeks to occur after positive tests, with another week or more until the deaths are actually reported as COVID-19 deaths. So there is a 3-5 lag between the case numbers and the death numbers. If we put a line in the cases graph we get:
Noting that the death numbers are small compared to the case numbers so the graph won't be a smooth, look again at the deaths graph above, compare that the part of the cases graph to the left of the line, then think about it a bit and tell me what seems likely to happen to deaths over the next month? 

If you look at the percentage increase in how many more tests they have done, the pattern of increase in cases follows the testing increase almost exactly. We don't have more cases, they are just testing more and detecting the asymptomatic cases that have always been there.

OK, first off we don't know the test positivity rate, so this just can't justify this claim. We do have an estimated R-nought value of 1.11 for Franklin County, meaning the problem is growing not shrinking. Overall in Ohio the test positivity rate has gone from 4.3% to 6.3% in the past month, which argues very strongly that this is not a case of just having increased testing rooting out asymptomatic cases that have been there all along. And the increases in all of Ohio outstrip the increases in testing. In fact about 1/3 of all the cases in Ohio have been identified in the past three weeks but nowhere near 1/3 of all the testing occurred in the corresponding time frame. 

Also, the Dublin area has had very low numbers all along, with only a few hundred cases total. Only about 0.4% of the population in Dublin has even had this. 

The number of cases is growing in Dublin, just like in the rest of Franklin County. This is an attempt to throw in numbers that sound like they might support the "this is not a serious problem" idea but they don't. Also, a large part of the Dublin City Schools is in the City of Columbus. So you really need to look at the number of cases in all the parts of DCS. Like Dublin, the cases in Columbus are going up rapidly.

The common flu also is transmitted asymptomatically, and is much more dangerous to children than coronavirus, yet we don't test everyone for the flu each year. 

and

Yes, vulnerable populations need to be protected, but shutting down businesses and schools and destroying our country is not the answer to this virus. I am glad the schools are giving an option for people who are vulnerable or are scared, but I really hope those of us that aren't are given an option for our kids to return to school full time.

The flu is not much more dangerous to kids than COVID-19, we do a ton of flu testing every year, we aren't testing anywhere near everyone for SAR-COV-2 (that's part of the problem), and kids are not the only people we are worried about. We are also worried about kids spreading the virus. We don't yet know whether COVID-19 is more deadly than common influenza's than kids. We are nowhere near being able to argue that. When we closed schools we isolated kids from infection. It does appear that younger kids (under 10) are much less likely to contract it than older kids and adults. So it might turn out to be less deadly. But so far so many fewer kids have contracted it that we just don't know. We do know that there are other serious health risks with COVID-19 that are far more common than in influenza. So the evidence right not points to COVID-19 not being less dangerous for kids than influenza, and certainly does not support a claim that is less dangerous. W

e do lots and lots of flu tests every year. including this year. We also have flu vaccines which allow us to mitigate risk and spread. Initial flu tests take tens of minutes, so they are done right in a doctor's office. If we had similar testing for COVID-19 we might have controlled it. 

Finally, the biggest worry is that kids will spread COVID-19 to school staff and to the their families and neighbors. Even if we knew no kid would get seriously ill, if we went back and did school as normal we would have a lot of teachers, bus drivers, custodians and cafeteria workers die in Dublin. Because almost all of the adults would get it. We would also have a really high percentage of the parents of our students get COVID-19, and some of the them would die. And even more of the adults would suffer long term health impairment. 



Some Links Used:









No comments: