[BostonCard]

I don't know where they get that 20% of cases in the US can be traced to Strugis.  The artice claims:

Finally, difference-in-differences (dose response) estimates show that following the Sturgis event, counties that contributed the highest inflows of rally attendees experienced a 7.0 to 12.5 percent increase in COVID-19 cases relative to counties that did not contribute inflows.

Basically, we are only talking about a 7 - 12.5% increase, and that is only in counties from which a lot of Sturgis participants came from.  In South Dakota, where the rally occurred, the effect was larger ("This represents an increase of over 35 percent relative to the 9.7 cases per 1,000 population in South Dakota on July 31, 2020 (South Dakota Department of Health 2020).") you would expect the effect to dissipate pretty quickly outside the state.

I guess I’m confused. The WaPo reports 260 cases, while saying that is low.  This study says a couple hundred thousand - based on what. Certainly not contact tracing  - pretty sure those at Sturgis aren’t going to be too inclined to submit.  So how did they reach a number, which on first blush, looks to be preposterous.  Also, I have to ask: do you consider  the Center for Health Economics & Policy Studies at San Diego State University, to be unbiased?

The difference is between direct and indirect effects.  The 260 represents the number of people who attended Sturgis and developed COVID-19 within the infectious period (~2 weeks) after the event.  The study by Stavia includes direct and indirect effects; not just the people directly infected with COVID-19, but all the people they spread it to.  From the paper (http://ftp.iza.org/dp13670.pdf):

Second, using data from the Centers for Disease Control and Prevention (CDC) and a synthetic control approach, we show that by September 2, a month following the onset of the Rally, COVID-19 cases increased by approximately 6 to 7 cases per 1,000 population in its home county of Meade. Finally, difference-in-differences (dose response) estimates show that following the Sturgis event, counties that contributed the highest inflows of rally attendees experienced a 7.0 to 12.5 percent increase in COVID-19 cases relative to counties that did not contribute inflows.

Note that the paper includes the corresponding author's contact information.  My experience is that authors are generally very willing to discuss their work, e-mail him and ask.  My own rather opinion based on a cursory examination is that there is nothing exotic or unusual about the methods the authors used (they approached it using the econometrics toolbox, not the public health toolbox; these are very similar, but you can tell from the fact that they did a "difference-in-difference" analysis that the authors were economists).  The results are subject to the same biases that any observational study (one where there isn't a randomized, placebo controlled intervention) are subject to.  As best I can tell, the authors made a good faithed effort to address the potential biases, but you can only address the biases you can think about, and there is always a possibility of residual confounding.

In my mind, the bottom line is that the study is solid and plausible but on its own should not be considered definitive.



BC

[BostonCard]

I don't know where they get that 20% of cases in the US can be traced to Strugis.  The artice claims:

Finally, difference-in-differences (dose response) estimates show that following the Sturgis event, counties that contributed the highest inflows of rally attendees experienced a 7.0 to 12.5 percent increase in COVID-19 cases relative to counties that did not contribute inflows.

Basically, we are only talking about a 7 - 12.5% increase, and that is only in counties from which a lot of Sturgis participants came from.  In South Dakota, where the rally occurred, the effect was larger ("This represents an increase of over 35 percent relative to the 9.7 cases per 1,000 population in South Dakota on July 31, 2020 (South Dakota Department of Health 2020).") you would expect the effect to dissipate pretty quickly outside the state.

I guess I’m confused. The WaPo reports 260 cases, while saying that is low.  This study says a couple hundred thousand - based on what. Certainly not contact tracing  - pretty sure those at Sturgis aren’t going to be too inclined to submit.  So how did they reach a number, which on first blush, looks to be preposterous.  Also, I have to ask: do you consider  the Center for Health Economics & Policy Studies at San Diego State University, to be unbiased?

The difference is between direct and indirect effects.  The 260 represents the number of people who attended Sturgis and developed COVID-19 within the infectious period (~2 weeks) after the event.  The study by Stavia includes direct and indirect effects; not just the people directly infected with COVID-19, but all the people they spread it to.  From the paper (http://ftp.iza.org/dp13670.pdf):

Second, using data from the Centers for Disease Control and Prevention (CDC) and a synthetic control approach, we show that by September 2, a month following the onset of the Rally, COVID-19 cases increased by approximately 6 to 7 cases per 1,000 population in its home county of Meade. Finally, difference-in-differences (dose response) estimates show that following the Sturgis event, counties that contributed the highest inflows of rally attendees experienced a 7.0 to 12.5 percent increase in COVID-19 cases relative to counties that did not contribute inflows.

Note that the paper includes the corresponding author's contact information.  My experience is that authors are generally very willing to discuss their work, e-mail him and ask.  My own rather opinion based on a cursory examination is that there is nothing exotic or unusual about the methods the authors used (they approached it using the econometrics toolbox, not the public health toolbox; these are very similar, but you can tell from the fact that they did a "difference-in-difference" analysis that the authors were economists).  The results are subject to the same biases that any observational study (one where there isn't a randomized, placebo controlled intervention) are subject to.  As best I can tell, the authors made a good faithed effort to address the potential biases, but you can only address the biases you can think about, and there is always a possibility of residual confounding.

In my mind, the bottom line is that the study is solid and plausible but on its own should not be considered definitive.



BC