04-08-2020, 10:41 AM
Not terribly surprising to me but it's interesting nonetheless: https://www.vox.com/2020/4/8/21212160/co...ion-deaths
The CardBoard
/
Emergency
/
Covid-19
/
Study: Small increases in air pollution make coronavirus much more deadly
Study: Small increases in air pollution make coronavirus much more deadly
Not terribly surprising to me but it's interesting nonetheless: https://www.vox.com/2020/4/8/21212160/co...ion-deaths
04-08-2020, 11:37 AM
I read this paper, and I don't doubt that air pollution makes thing worse. But this analysis has a fatal flaw: rural counties tend to have better air quality all things being equal: https://www.cdc.gov/mmwr/volumes/66/ss/ss6613a1.htm And rural counties were later than urban areas to get the virus, so their death rates will lag: https://www.nytimes.com/interactive/2020...cases.html I suspect you can probably explain much of the differences between areas in deaths by population.
This paper is not published yet and shouldn't be. I think we're going to see a lot of retractions and/or fatal critiques of papers that are being put out now.
Someone should go back and do this analysis once we have final death rates and after accounting for the timing of when social distancing measures were implemented and the timing of the first case in each county.
This paper is not published yet and shouldn't be. I think we're going to see a lot of retractions and/or fatal critiques of papers that are being put out now.
Someone should go back and do this analysis once we have final death rates and after accounting for the timing of when social distancing measures were implemented and the timing of the first case in each county.
I read this paper, and I don't doubt that air pollution makes thing worse. But this analysis has a fatal flaw: rural counties tend to have better air quality all things being equal: https://www.cdc.gov/mmwr/volumes/66/ss/ss6613a1.htm And rural counties were later than urban areas to get the virus, so their death rates will lag: https://www.nytimes.com/interactive/2020...cases.html I suspect you can probably explain much of the differences between areas in deaths by population.
This paper is not published yet and shouldn't be. I think we're going to see a lot of retractions and/or fatal critiques of papers that are being put out now.
Someone should go back and do this analysis once we have final death rates and after accounting for the timing of when social distancing measures were implemented and the timing of the first case in each county.
This paper is not published yet and shouldn't be. I think we're going to see a lot of retractions and/or fatal critiques of papers that are being put out now.
Someone should go back and do this analysis once we have final death rates and after accounting for the timing of when social distancing measures were implemented and the timing of the first case in each county.
04-08-2020, 12:30 PM
From the original paper, "We adjust by population size, hospital beds, number of individuals tested, weather, and socioeconomic and behavioral variables including, but not limited to obesity and smoking. We include a random intercept by state to account for potential correlation in counties within the same state."
The possibility of residual confounding cannot be discounted, however.
BC
The possibility of residual confounding cannot be discounted, however.
BC
From the original paper, "We adjust by population size, hospital beds, number of individuals tested, weather, and socioeconomic and behavioral variables including, but not limited to obesity and smoking. We include a random intercept by state to account for potential correlation in counties within the same state."
The possibility of residual confounding cannot be discounted, however.
BC
The possibility of residual confounding cannot be discounted, however.
BC
(04-08-2020, 12:30 PM)BostonCard Wrote: From the original paper, "We adjust by population size, hospital beds, number of individuals tested, weather, and socioeconomic and behavioral variables including, but not limited to obesity and smoking. We include a random intercept by state to account for potential correlation in counties within the same state."
The possibility of residual confounding cannot be discounted, however.
BC
Just dug in the supplementary materials. When they excluded counties with <10 cases (those should mostly be rural, but i haven't downloaded the data to look), they got non-significant results for 3 of 4 tests. The population size adjustment did not change the results, which is contrary to what I expected. I'd really want to dig into the data to understand the interaction of population size and death rate.
Anyway, I still smell a rat here. Overall effect size is small and P values are ~0.01 despite a brobdingnagian sample size of counties. Also, they didn't adjust for spatial autocorrelation. Autocorrelation in pollution is obvious from their maps, so I would bet they have spatial autocorrelation in their residuals. If so, that would violate model assumptions and call the results into question. If I get some time (unlikely), I'll download the repo and look at the residuals.
If I was a reviewer on this paper, I would send it back.
(04-08-2020, 12:30 PM)BostonCard Wrote: From the original paper, "We adjust by population size, hospital beds, number of individuals tested, weather, and socioeconomic and behavioral variables including, but not limited to obesity and smoking. We include a random intercept by state to account for potential correlation in counties within the same state."
The possibility of residual confounding cannot be discounted, however.
BC
Just dug in the supplementary materials. When they excluded counties with <10 cases (those should mostly be rural, but i haven't downloaded the data to look), they got non-significant results for 3 of 4 tests. The population size adjustment did not change the results, which is contrary to what I expected. I'd really want to dig into the data to understand the interaction of population size and death rate.
Anyway, I still smell a rat here. Overall effect size is small and P values are ~0.01 despite a brobdingnagian sample size of counties. Also, they didn't adjust for spatial autocorrelation. Autocorrelation in pollution is obvious from their maps, so I would bet they have spatial autocorrelation in their residuals. If so, that would violate model assumptions and call the results into question. If I get some time (unlikely), I'll download the repo and look at the residuals.
If I was a reviewer on this paper, I would send it back.
04-08-2020, 01:25 PM
I wonder if there is any correlation between NYC Covid-19 deaths and the dust from the 9/11 twin towers attacks affecting the lungs.
Put a glide in your stride and a dip in your hip.
04-08-2020, 01:40 PM
(04-08-2020, 01:25 PM)martyup Wrote: I wonder if there is any correlation between NYC Covid-19 deaths and the dust from the 9/11 twin towers attacks affecting the lungs.And, in Italy, the dust from WWII affects the 75+yo group! (;-)
Oh yea, Iran had it bad too.
I had seen the pollution paper and decided it wasn't worth my time. I feel a bit bad dismissing it out of hand, but there is only so much time. I'm glad others will evaluate it more properly.
(04-08-2020, 01:25 PM)martyup Wrote: I wonder if there is any correlation between NYC Covid-19 deaths and the dust from the 9/11 twin towers attacks affecting the lungs.And, in Italy, the dust from WWII affects the 75+yo group! (;-)
Oh yea, Iran had it bad too.
I had seen the pollution paper and decided it wasn't worth my time. I feel a bit bad dismissing it out of hand, but there is only so much time. I'm glad others will evaluate it more properly.
04-08-2020, 01:51 PM
I was skeptical of the pollution study as well. Could be the case, but way too early to have the data you'd need to make that argument.
I suspect we will know with good data how the 9/11 responders do with this. They are a very well studied cohort. But again, we will need to wait a bit longer to have anything we can support with good data.
I suspect we will know with good data how the 9/11 responders do with this. They are a very well studied cohort. But again, we will need to wait a bit longer to have anything we can support with good data.
I was skeptical of the pollution study as well. Could be the case, but way too early to have the data you'd need to make that argument.
I suspect we will know with good data how the 9/11 responders do with this. They are a very well studied cohort. But again, we will need to wait a bit longer to have anything we can support with good data.
I suspect we will know with good data how the 9/11 responders do with this. They are a very well studied cohort. But again, we will need to wait a bit longer to have anything we can support with good data.
05-08-2020, 10:24 AM
Okay, here's a much much better study of air pollution effects on mortality in the US: https://www.medrxiv.org/content/10.1101/...20090746v1
tl,dr: NO2 (nitrogen dioxide) was associated with moderately higher death rates. I.e., going from the average NO2 level in the US to the top 25 percentile would increase death rates about 7% (that's percent, not percentage points). Ozone and PM2.5 had no effects.
This study dotted all their i's, etc. I just skimmed it, but I don't see any major flaws.
NO2 tends to be highest near big cities.
tl,dr: NO2 (nitrogen dioxide) was associated with moderately higher death rates. I.e., going from the average NO2 level in the US to the top 25 percentile would increase death rates about 7% (that's percent, not percentage points). Ozone and PM2.5 had no effects.
This study dotted all their i's, etc. I just skimmed it, but I don't see any major flaws.
NO2 tends to be highest near big cities.
Okay, here's a much much better study of air pollution effects on mortality in the US: https://www.medrxiv.org/content/10.1101/...20090746v1
tl,dr: NO2 (nitrogen dioxide) was associated with moderately higher death rates. I.e., going from the average NO2 level in the US to the top 25 percentile would increase death rates about 7% (that's percent, not percentage points). Ozone and PM2.5 had no effects.
This study dotted all their i's, etc. I just skimmed it, but I don't see any major flaws.
NO2 tends to be highest near big cities.
tl,dr: NO2 (nitrogen dioxide) was associated with moderately higher death rates. I.e., going from the average NO2 level in the US to the top 25 percentile would increase death rates about 7% (that's percent, not percentage points). Ozone and PM2.5 had no effects.
This study dotted all their i's, etc. I just skimmed it, but I don't see any major flaws.
NO2 tends to be highest near big cities.
05-08-2020, 10:43 AM
Isn't there a difference between those exposed to air pollution developing asthma that makes dealing with Covid-19 more difficult, and Covid-19 being more difficult to deal with in polluted air for someone without asthma? Correlation vs. causation?
Isn't there a difference between those exposed to air pollution developing asthma that makes dealing with Covid-19 more difficult, and Covid-19 being more difficult to deal with in polluted air for someone without asthma? Correlation vs. causation?
05-08-2020, 12:11 PM
(05-08-2020, 10:43 AM)OutsiderFan Wrote: Isn't there a difference between those exposed to air pollution developing asthma that makes dealing with Covid-19 more difficult, and Covid-19 being more difficult to deal with in polluted air for someone without asthma? Correlation vs. causation?
It turns out that asthma is not an overwhelming risk factor for worsening COVID-19, if you compare the proportion of COVID-19 hospitalized patients with asthma, and the proportion of patients who die from COVID-19 with asthma to the local prevalence of asthma. The data is not precise enough to exclude a small effect, but compared to other risk factors (diabetes, hypertension), it is not whopping.
BC
(05-08-2020, 10:43 AM)OutsiderFan Wrote: Isn't there a difference between those exposed to air pollution developing asthma that makes dealing with Covid-19 more difficult, and Covid-19 being more difficult to deal with in polluted air for someone without asthma? Correlation vs. causation?
It turns out that asthma is not an overwhelming risk factor for worsening COVID-19, if you compare the proportion of COVID-19 hospitalized patients with asthma, and the proportion of patients who die from COVID-19 with asthma to the local prevalence of asthma. The data is not precise enough to exclude a small effect, but compared to other risk factors (diabetes, hypertension), it is not whopping.
BC
« Next Oldest | Next Newest »