The Sturgis Super-spreader Study:
There have been some misunderstandings about this study, so I want to first apologize for any confusion caused by my writing. I value the trust of readers and endeavor to be worthy of it. My overarching conclusion that mask-wearing can slow the spread of this virus and has very little for downsides remains valid.
Miscommunication is never the fault of the public, and for that I take responsibility. While the rally almost certainly increased spread, no one can say how many people caught the virus at the event. More complicated downstream effects would be even less certain. That should have been more clearly expressed.
To be clear, blaming members of the public for attending is inappropriate regardless. Assigning fault or blame is not the place of science. That is not something that could be inferred from a study. More importantly, if blame were assigned, it must fall to leadership that willfully accepted the role of informing and safeguarding people. That is the burden and privilege of public service.
It is impossible to say anything with certainty concerning the number of people infected by the virus as a consequence of the Rally. This is an economics paper looking at infectious disease, so we should view the results with skepticism. This is merely a presentation of the conclusions of one study, and we must see those conclusions as uncertain.
“This study, which was available as a preprint and thus had not yet been peer-reviewed, uses county-level SARS-CoV-2 testing data to show that the Sturgis Motorcycle Rally likely led to substantial increases in cases in the local community where the rally took place.
However, there is considerable uncertainty surrounding the broader national impact of the rally and its associated costs, given limitations in the methodological approaches used. Results from this study should be interpreted cautiously.”
Between August 7 and August 16, 2020, nearly 500,000 motorcyclists rolled their way into South Dakota for the 80th annual Sturgis Motorcycle Rally. The potential for a super-spreader incident goes without saying.
A non-peer-reviewed study by economists reported the “COVID-19 mitigation efforts at the Sturgis Rally were largely left to the ‘personal responsibility’ of attendees, and post-opening day media reports suggest that social distancing and mask-wearing were rare in Sturgis.”
The Sturgis Motorcycle Rally likely increased spread, but how much is something we cannot say with certainty. A recent study estimated 267,000 more people between August 2 and September 2 of 2020 could be attributed to the virus.
Considering the minuscule percentage of Americans in attendance, that is an exceptionally large contribution and more evidence is needed to say anything with certainty — but the good news would be that Sturgis shows us we have a choice about whether big events lead to steep increases in case counts.
The jury is still out on the truth of the matter but as the discussion progresses pay more attention to general themes, we find than specific numbers. Do we see trends up or down? How does that fit with the other evidence?
To put it simply, a lot of math. Here is the formula for calculating, “Effect of Sturgis Rally on Non-Resident Travel and Foot Traffic in Sturgis” in case you wanted a headache:
Looking at the number of people traveling to Sturgis, expected rates of spread before the rally, the policies in place in home states, and understanding the inflow and outflow of people, all played a role in this estimate. For a more detailed explanation, see the study.
Nearly half a million people passed through during the 10-day event, which takes place in a town of just 7,000 people. The tremendous difference between the normal population and the numbers seen during Sturgis paints a picture of tightly cramped spaces. Using cell phone towers as a proxy for where people are and when cramped is an appropriate description.
How much the Covid cases increased when people returned home, appears to be “dose-dependent.” The more bikers coming from a single location, the greater the risk for increased spread when they returned home. Three weeks after the start of Sturgis, case counts rose nearly 15% in certain states.
Where cases rose tells us as much as where they did not. Certain locations that contributed large numbers of attendees did not see any real increase. How could that be?
Two factors appear to play critical roles in what happens when people return home from Sturgis:
Areas that had few residents travel to Sturgis had little to no increase, but that wasn’t the most interesting find. States with strong policies in place, namely mask mandates, saw much lower Covid-19 case counts in the week after Sturgis ended. That trend matters more than the specific numbers.
Areas with a mask mandate saw such limited increases in case numbers — even if many residents had traveled to Sturgis — that the study authors reported it was not statistically significant. Put another way, even if a huge number of bikers from a single location returned home carrying Covid-19, whether or not the travelers triggered a climb in cases depended on the policies in place.
The returning travelers did not cause explosive increases in cases if the state enacted measures to control the spread, namely mask-wearing. That’s incredible.
It’s hard to overstate the power masks have to return us to a functional society. Recent and hopeful evidence shows universal mask-wearing may have other benefits as well. That’s a reason to have hope. Even something that could be the largest documented super-spreader incident in recorded history appears not to have been a match for a state filled with mask-wearers. It’s also worth noting if we’re wrong about masks in some way, there’s little in the way of downsides.
The protective effect of masks supports the conclusions drawn by the Wall-Street Journal and Goldman-Sachs that mask-wearing could protect our economy and the economy will not recover until we control the outbreak. Wall-Street concluded that the economy would not recover without an effective response. Goldman-Sachs reported that masks could stand-in for a shutdown and reduce the long-term damage to the economy.
With BLM protests and political rally in Tulsa, local populations mingled less with the crowds. The risk increases that the crowded events might have caused appear to be and have prompted locals to take more precautions. Sturgis saw the opposite.
Residents mingled with visitors at Sturgis and stopped stay-at-home behavior. Researchers saw this using cell phone tower data. Attendees also traveled significantly farther where protests tended to be more localized. Critically, as this study shows, states that had many people attend Sturgis did not see cases case counts climb if they had mitigation measures in place.
Expecting cases to increase following a protest seems logical. What we failed to account for was the response of the local communities. Locations with protests saw residents stay home more and take far more precautions than they did prior to the event.
Researchers graphed locations with protests and considered aspects like media coverage, violence, size of the protest, how quickly the virus was spreading where they were before the protest, whether people took steps to stop the spread, and the length of protests, among other factors.
The blue lines represent what we’d expect without a protest based on what was happening before they took place. The red line shows what happened; the results are striking.
Protests also took place outdoors, and the virus has shown itself not to spread well outdoors. That paired with the response of locals may have been enough to prevent spikes. Behavior and people determine how much the virus costs their communities and the state.
Much of Sturgis took place outdoors too, but restaurants, bars, and other sheltered settings laid the groundwork for massive spread. If people wore masks at Sturgis or if attendees’ home-states had a mask policy, the spread did not lead to case increases.
The costs incurred by the extra cases will disproportionately affect locations without measures because that is where the burden and the sick will concentrate. People tend to focus on the obvious costs while forgetting that there are many real costs that can be difficult to measure in the immediate aftermath. That controlling spread helps the economy falls in line with conclusions drawn by an MIT study in March 2020.
The authors assume the implausible survival of every single Sturgis case and used a prior study’s estimates on the cost per case to estimate a cost of $12.2 billion dollars. Regardless of how someone became infected, 1.4 million Americans tested positive and there are costs associated with that.
If that amount were true, that comes to $26,553.64 for each of the 462,182 persons who attended Sturgis. Should this estimate reflect reality, we could have paid Sturgis attendees that amount per person to stay home and come out ahead.
It’s worth noting that the cost of getting sick often far exceeds what we see on the surface. That tendency frequently causes leaders to underestimate the cost of not safeguarding health.
A deep-state cabal — just kidding. Rather than “coming up” with a number, economists, statisticians, epidemiologists, and several other fields detect and measure the impact a disease has on our lives. What we must be careful to note is that the scientific process is taking place in public where normally ideas would not see the public until we had finished critiquing each others’ work. That limits the amount of incorrect information that goes to the public. The nature of this crisis is difficult in part because of that public stage to the process.
Having said that, the costs of infectious disease often exceed past income or the cost for treatment, but it’s also not immediately obvious. When the World Bank estimated that a moderate pandemic would cost the world at least $3 trillion dollars, people scoffed at it. Turns out, they were right, conservative even.
If a person catches the seasonal flu, people usually consider the lost wages or hospitalization. That misses critical costs like those that build up over time. In 2018 a study found that the risk of heart attack increased six timeshigher than the norm within a week of influenza infection.
If you’re only considering the lost employee hours or the death rate, you’re missing what is sometimes the bigger picture. Polio, another disease that spreads asymptomatically, has rare but serious consequences in a minority of cases. Just considering those who die is to forget President Franklin Delano Roosevelt. There is a wide spectrum between perfect health and death.
In the case of the flu example, a heart attack can easily shorten life expectancy, but what about those who have a less apparent injury? What sort of costs would come later in life for someone with permanent lung damage for example? Health is hard to beat in cost-effectiveness and we’re misleading ourselves if we think we’re saving money by leaving people to develop advanced, preventable diseases.
Some cases have a relatively low cost, while others will require hospitalization. All-inclusive costs often consider and factor in expenses that people rarely consider when estimating the cost of illness, or the burden of disease as academics sometimes call it.
When people catch Covid-19, there are intangible costs we know exist from studying the impact of other diseases. What we’ve learned over time is that illness and death (though we’re not considering death in this calculation) have costs that far exceed the obvious expenses. Americans think of the world transactionally, not seeing downstream costs that extend beneath the surface.
If you’re a leader considering the costs of an influenza outbreak, you must consider the downstream effects. Which would you consider? How and why? What formula would you apply to the specific population in your state? How could you cross-check it to see if it’s accurate?
Here’s an example that shows the idea more clearly. Often when people and communities consider contributing the education of a woman who intends to stay home with her children, people discuss it in terms of whether the education allows her to bring in income. That’s one benefit but we often fail to consider what may be the more important benefits, like better health outcomes for her and children, lower risk for abuse, lower risk for criminal activity, better health across a lifetime, and longer lives.
Women tend to invest 90% of what they earn into their families giving their children advantages that cannot be gotten later in life, and on a national scale, “one percentage point increase in female education raises the average gross domestic product (GDP) by 0.3 percentage points and raises annual GDP growth rates by 0.2 percentage points.”
It’s a simple decision in terms of whether it’s worth it for the government to pay for part or all of someone’s education. Like most public health measures, it works for both parties.
Who we educate affects society on the whole. Literally, I benefit from the education of those I have never met. It’s why many other countries have made education either more affordable or funded in the form of taxes. The US is unfortunately poor in these kinds of assessments. Even when decision-makers understand it, it can be a hard sell with voters.
Another example: Cutting our pandemic prevention and prediction program PREDICT in October 2019 was illogical in terms of risk and fiscally irresponsible. You would be hard-pressed to find an economist or insurance agency that would recommend you end the program, yet we did. Why?
It probably felt like savings at the time. The program cost 1/143,000 of the expected cost of a single pandemic, and the project paid for itself via the information gained while trying to prevent spillover. What other initiatives have a return on investment? Very few. Public health measures are paradoxically difficult to get funded while also being one of the most likely to pay for itself.