D Hart, these are two poorly designed studies which found what they set out to find. There are around 150 more like them. It's a very strange time for science. Quick review of the two you linked:
"Mask Effectiveness for Preventing Secondary Cases of COVID-19, Johnson County, Iowa,"
1) It's a retrospective case control study done through questionnaires, not RCT, so at bottom tier of journal quality.
2) They estimated 50% difference, did retrospective case control study and found 50% difference. What are the odds?
3) Even accepting the design flaws, if you read the data in table 1 (https://wwwnc.cdc.gov/eid/article/28/1/21-1591-t1) their findings show the best odds of avoiding Covid is if Index patient does NOT wear a mask. This was 20% lower than if index wore a mask.
4) Setting aside this study, Iowa had among the lowest mask use for students, and despite claims in the media like "Welcome to Iowa, a state that doesn’t care if you live or die"[1] they fared no different than anyone else, again, driving home the point that despite what we hoped, very little (if any) of our mitigation strategies had any demonstrable benefits without running them a case control study so we can data drudge and p-hack.
For the other study:
"Effectiveness of Face Mask or Respirator Use in Indoor Public Settings for Prevention of SARS-CoV-2 Infection — California, February–December 2021"
1) This study was lampooned so hard by left, right, and middle on mask debate that the author had to make her twitter bio private after pushback (which is sad, I wish people would be more civil). But seriously, that is a terrible study.
2) I will let very pro mask Emily Oster refute this one:
Regarding your #3 and the Iowa study, the results do not show what you state. Due to the small numbers in the unmasked/masked group, and in the masked/unmasked group, the confidence intervals are wide (about 3X wider than those for the unmasked/unmasked and masked/masked groups).
The CIs for the unmasked/masked group and the masked/unmasked group (which are the groups I assume you're comparing) overlap. This means that the data presented does not show a difference between the 2 groups. There may be a difference (or there may not be), but the data presented did not show it. It certainly does not "show the best odds of avoiding Covid is if Index patient does NOT wear a mask." You are misinterpreting the results, though it is a common mistake.
To get a better idea of the effectiveness of masks using the CDC data for the Iowa study, examine the data (and, importantly, the CIs) for the 2 groups with the much greater number of cases (more than 9X) relative to the discordant groups.
For those in which both were masked, the results were 12.5 (9.6–16.3); for those in which both were unmasked, the results were 26.4 (22.9– 30.7). Note that the upper limit of the secondary attack rate for those in which both were masked was 16.3, significantly lower than the lower limit of the secondary attack rate for those in which both were unmasked (22.9), and there was no overlap in the CIs. This is consistent with the hypothesis that masking significantly reduces the secondary attack rate.
Emily has been consistently pro mask throughout her writings the last 18 months, though her enthusiasm hasn't been as pronounced probably since Omicron (this is just personal gauge of reading her substack and previously her twitter feed).
Regardless, careful falling to the logical fallacy of "ad hominen" - you are attacking the arguer (she doesn't have the credentials) not the argument itself (the study is poorly designed). By that logic should we have dismissed Katelyn writings when she rose to prominence on social media just because prior to 2020 her published papers were largely focused on domestic violence, poverty, LE, economic stressors, etc, and not on infectious disease?
Regarding the study you linked: "Mask adherence and rate of COVID-19 across the United States", the study reference period was Apr - Sept 2020, then compare to May - Oct 2020.
All claims it makes should you re-run for future dates fall apart. There are a number of subsequent studies that make a similar mistake, not realizing that the seasonality would bias results against warmer climate states when restricting to that time frame.
For reference there 700K covid cases in California during the study reference period. That number would triple within 4 months, despite California having some the strictest mask mandates and adoption rates. Or use Illinois, it *only* had 236K cases during the reference period, which is credited towards masks. But within 4 months, cases quadrupled. But in Florida, cases only doubled between Sept - Dec. Was this due to masks then?
Studies like this, if their claims are falsifiable, should produce the same results when date ranges are adjusted (i.e., you should be able to reproduce the claims regardless of the 5 month window of time you choose). In this case, it fails that replication test.
KB: "So, let's be clear here. You appealed to authority in using a link to Emily's and I merely pointed out she is a professor of Economics."
MD: There is no appeal to authority, I only noted that Emily is someone who believes in masks, so for her to pick apart a pro-mask study is notable. I figured linking her rebuttal as opposed to Vinay Prasad would reduce risk of dismissal by ad hominem.
I'm familiar with positive and negative study design, though hadn't seen that link, which is useful as a refresher.
As for the rest of your post - I'm not entirely clear on your point, it seems like you are re-refuting the John Hopkins study Emily also picked apart, is that your intention? If so, I don't disagree with either her or your critique of that paper, even if the point it establishes is probably true. But again, seems rather than tackling the arguments you are focused on the credentials and political affiliations of the authors, which goes back to my original point - attack the argument not the arguer (and at this point, I think Sweden is vindicated in their approach).
KB - I reached that conclusion by comparing the excess mortality in Sweden against her Nordic neighbors, Europe, Israel, the US, and South Korea.
Evaluating excess deaths for 2020-2021, Sweden had 4.2% more deaths than usual (using 2016-2019 as baseline), which ranks below all of Europe (range of 8%-20%), South Korea (8.8%), Israel (12%), the United States (21%), slightly below Finland (5.5%) and Denmark (4.9%). Only Norway (2%) had lower excess deaths. [1]
Additionally, when you examine the details of the excess deaths, Sweden had zero excess deaths in the 0-64 age group in 2020 and 2021. [2]
Note that the article you linked was to the preliminary findings of the commission from October. In the final report recently published the tone was softer: "In comparison with the rest of Europe, Sweden has come through the pandemic relatively well and is among the countries with the lowest excess mortality over the period 2020-2021". [3] While I don't put much stock in the report either way - politicians are prone to saying they were right while always leaving window open to "we could have done better" - I suspect the softer tone was due to Denmark exploding in cases in the end of 2021 and early 2022, but just speculating.
And finally, I think there has been a double standard regarding Sweden - critiques always compare them only to their Nordic neighbors, never to Germany, Israel, the UK, etc. Which is strange because we don't hold restrictions of comparison to anyone else. We compare the US to South Korea, Japan, New Zealand all the time, yet Stockholm is the same distance to Berlin as Chicago is to Denver.
Yet the comparisons to Finland, Denmark, and Norway, even we accept them, fail to realize that all of those countries sent their kids back to school in the spring of 2020 without masks anyway, and had the lowest mask adoption rates in the entire world - which makes it odd to cite them as "doing it right" when arguing masks made a measurable difference in the pandemic.
_________
Sources:
I made this graphic a month ago, it compiles the sources below, but is slightly out of date (i.e., Finland since then added another 1,000 deaths to 2021), but it gives good visual of what I am saying if you don't want to bother recreating pivot tables https://imgur.com/a/EpbGWbg
In IL DCFS stated there’s no mask mandate anymore for daycare & preschool. Unreal
There’s been no masks in over half of the US for months, and no different outcomes than those with masks.
Alternatively just look at many European countries which haven’t had masks on children throughout the pandemic (especially Nordic countries).
You will see it will be ok, just as it was in all those places.
"There’s been no masks in over half of the US for months, and no different outcomes than those with masks."
That's not what these 2 studies showed.
https://wwwnc.cdc.gov/eid/article/28/1/21-1591_article
https://www.cdc.gov/mmwr/volumes/71/wr/mm7106e1.htm?s_cid=mm7106e1_w%20[cdc.gov]
D Hart, these are two poorly designed studies which found what they set out to find. There are around 150 more like them. It's a very strange time for science. Quick review of the two you linked:
"Mask Effectiveness for Preventing Secondary Cases of COVID-19, Johnson County, Iowa,"
1) It's a retrospective case control study done through questionnaires, not RCT, so at bottom tier of journal quality.
2) They estimated 50% difference, did retrospective case control study and found 50% difference. What are the odds?
3) Even accepting the design flaws, if you read the data in table 1 (https://wwwnc.cdc.gov/eid/article/28/1/21-1591-t1) their findings show the best odds of avoiding Covid is if Index patient does NOT wear a mask. This was 20% lower than if index wore a mask.
4) Setting aside this study, Iowa had among the lowest mask use for students, and despite claims in the media like "Welcome to Iowa, a state that doesn’t care if you live or die"[1] they fared no different than anyone else, again, driving home the point that despite what we hoped, very little (if any) of our mitigation strategies had any demonstrable benefits without running them a case control study so we can data drudge and p-hack.
For the other study:
"Effectiveness of Face Mask or Respirator Use in Indoor Public Settings for Prevention of SARS-CoV-2 Infection — California, February–December 2021"
1) This study was lampooned so hard by left, right, and middle on mask debate that the author had to make her twitter bio private after pushback (which is sad, I wish people would be more civil). But seriously, that is a terrible study.
2) I will let very pro mask Emily Oster refute this one:
https://emilyoster.substack.com/p/lots-of-studies-are-bad?s=r
____________
[1] https://www.washingtonpost.com/outlook/2021/02/10/iowa-lift-all-restrictions/
Thanks for your expansive reply.
Regarding your #3 and the Iowa study, the results do not show what you state. Due to the small numbers in the unmasked/masked group, and in the masked/unmasked group, the confidence intervals are wide (about 3X wider than those for the unmasked/unmasked and masked/masked groups).
The CIs for the unmasked/masked group and the masked/unmasked group (which are the groups I assume you're comparing) overlap. This means that the data presented does not show a difference between the 2 groups. There may be a difference (or there may not be), but the data presented did not show it. It certainly does not "show the best odds of avoiding Covid is if Index patient does NOT wear a mask." You are misinterpreting the results, though it is a common mistake.
To get a better idea of the effectiveness of masks using the CDC data for the Iowa study, examine the data (and, importantly, the CIs) for the 2 groups with the much greater number of cases (more than 9X) relative to the discordant groups.
For those in which both were masked, the results were 12.5 (9.6–16.3); for those in which both were unmasked, the results were 26.4 (22.9– 30.7). Note that the upper limit of the secondary attack rate for those in which both were masked was 16.3, significantly lower than the lower limit of the secondary attack rate for those in which both were unmasked (22.9), and there was no overlap in the CIs. This is consistent with the hypothesis that masking significantly reduces the secondary attack rate.
No, thank you! I rarely get pushback when diving into these studies. Appreciate your thoughtful feedback!
Emily has been consistently pro mask throughout her writings the last 18 months, though her enthusiasm hasn't been as pronounced probably since Omicron (this is just personal gauge of reading her substack and previously her twitter feed).
Regardless, careful falling to the logical fallacy of "ad hominen" - you are attacking the arguer (she doesn't have the credentials) not the argument itself (the study is poorly designed). By that logic should we have dismissed Katelyn writings when she rose to prominence on social media just because prior to 2020 her published papers were largely focused on domestic violence, poverty, LE, economic stressors, etc, and not on infectious disease?
Regarding the study you linked: "Mask adherence and rate of COVID-19 across the United States", the study reference period was Apr - Sept 2020, then compare to May - Oct 2020.
All claims it makes should you re-run for future dates fall apart. There are a number of subsequent studies that make a similar mistake, not realizing that the seasonality would bias results against warmer climate states when restricting to that time frame.
For reference there 700K covid cases in California during the study reference period. That number would triple within 4 months, despite California having some the strictest mask mandates and adoption rates. Or use Illinois, it *only* had 236K cases during the reference period, which is credited towards masks. But within 4 months, cases quadrupled. But in Florida, cases only doubled between Sept - Dec. Was this due to masks then?
Studies like this, if their claims are falsifiable, should produce the same results when date ranges are adjusted (i.e., you should be able to reproduce the claims regardless of the 5 month window of time you choose). In this case, it fails that replication test.
KB: "So, let's be clear here. You appealed to authority in using a link to Emily's and I merely pointed out she is a professor of Economics."
MD: There is no appeal to authority, I only noted that Emily is someone who believes in masks, so for her to pick apart a pro-mask study is notable. I figured linking her rebuttal as opposed to Vinay Prasad would reduce risk of dismissal by ad hominem.
I'm familiar with positive and negative study design, though hadn't seen that link, which is useful as a refresher.
As for the rest of your post - I'm not entirely clear on your point, it seems like you are re-refuting the John Hopkins study Emily also picked apart, is that your intention? If so, I don't disagree with either her or your critique of that paper, even if the point it establishes is probably true. But again, seems rather than tackling the arguments you are focused on the credentials and political affiliations of the authors, which goes back to my original point - attack the argument not the arguer (and at this point, I think Sweden is vindicated in their approach).
KB - I reached that conclusion by comparing the excess mortality in Sweden against her Nordic neighbors, Europe, Israel, the US, and South Korea.
Evaluating excess deaths for 2020-2021, Sweden had 4.2% more deaths than usual (using 2016-2019 as baseline), which ranks below all of Europe (range of 8%-20%), South Korea (8.8%), Israel (12%), the United States (21%), slightly below Finland (5.5%) and Denmark (4.9%). Only Norway (2%) had lower excess deaths. [1]
Additionally, when you examine the details of the excess deaths, Sweden had zero excess deaths in the 0-64 age group in 2020 and 2021. [2]
Note that the article you linked was to the preliminary findings of the commission from October. In the final report recently published the tone was softer: "In comparison with the rest of Europe, Sweden has come through the pandemic relatively well and is among the countries with the lowest excess mortality over the period 2020-2021". [3] While I don't put much stock in the report either way - politicians are prone to saying they were right while always leaving window open to "we could have done better" - I suspect the softer tone was due to Denmark exploding in cases in the end of 2021 and early 2022, but just speculating.
And finally, I think there has been a double standard regarding Sweden - critiques always compare them only to their Nordic neighbors, never to Germany, Israel, the UK, etc. Which is strange because we don't hold restrictions of comparison to anyone else. We compare the US to South Korea, Japan, New Zealand all the time, yet Stockholm is the same distance to Berlin as Chicago is to Denver.
Yet the comparisons to Finland, Denmark, and Norway, even we accept them, fail to realize that all of those countries sent their kids back to school in the spring of 2020 without masks anyway, and had the lowest mask adoption rates in the entire world - which makes it odd to cite them as "doing it right" when arguing masks made a measurable difference in the pandemic.
_________
Sources:
I made this graphic a month ago, it compiles the sources below, but is slightly out of date (i.e., Finland since then added another 1,000 deaths to 2021), but it gives good visual of what I am saying if you don't want to bother recreating pivot tables https://imgur.com/a/EpbGWbg
[1] https://www.mortality.org/
[2] Note - this is a direct download xls file: https://www.scb.se/en/finding-statistics/statistics-by-subject-area/population/population-composition/population-statistics/pong/tables-and-graphs/preliminary-statistics-on-deaths/
[3] https://www.thelocal.se/20220225/swedens-pandemic-strategy-fundamentally-correct-coronavirus-commission/
Yes