The Sturgis Motorcycle Rally took place from Aug 7 to Aug 16 in Sturgis, Meade County, South Dakota. Some 60% of the local residents feared the worst and voted against holding the rally in 2020. However, the City Council and the Mayor, supported by the Governor, Kristi Noem (R), decided to go ahead anyway, primarily since they projected that they would make a few million dollars in city and state tax revenue.
Epidemiologists warned that the cost of this “superspreader” could be high. Virtually no coronavirus precautions were taken during the rally, which drew some 450,000 people from almost all the states in the continental US, from an estimated 61% of the counties. Unlike the 29 states which continue to ban gatherings of over 50 individuals, South Dakota bases their coronavirus strategy on the “honor system” In Sturgis, attendees huddled together in local bars, tattoo parlors, restaurants, shops, campgrounds, and enjoyed live music of rock bands, races, poker tournaments, bike shows and other more raucous adult entertainment in a Wild West kind of atmosphere.
Now the results are coming in. We know that the number of coronavirus cases that can be directly traced to Sturgis has surpassed 260, according to the data published by the Departments of Health of 12 states. Last week, Minnesota DOH reported the first confirmed COVID-19 death arising from the Sturgis Rally. “He was a man in his 60s and had underlying health conditions; he had been hospitalized for several weeks.”
Well, 260 seems incredibly small. Indeed, if this number were to represent the stable state of COVID-19 spread from such a reckless concentration of people in the times of COVID, we would be justified in supposing that schools and businesses could be immediately opened with a modicum of precautions. Wouldn’t that be nice?
Unfortunately, 260 is quite literally incredibly small. Thus far, State Departments of Health have relied exclusively on self-reporting to link local cases to Sturgis. Since the self-reporting is not even mandatory, one suspects that it has been done very rarely. Contact tracing of COVID-19 cases has always been problematic, and the 450,000 attendees undoubtedly include a significant group of asymptomatic spreaders to non-attendees, further complicating any effective, um, postmortem analysis.
It only takes a casual glance at the numbers from the DOH of South Dakota, the “mother state,” to realize that something is way off. Officially, South Dakota has just a hair more than 100 cases from Sturgis. (Minnesota, the second most Sturgis-affected state attributes only 50 to the rally.) However, South Dakota as a whole has seen rapidly increasing coronavirus cases since the rally. Last week alone it reported more than 2,000 new cases, setting single day records several times. Meade County, where Sturgis is located, had 71 COVID-19 cases before the start of the rally, on August 1; today, that number had risen sharply to 346.
There seems to be no direct way to get the true number of Sturgis-linked coronavirus cases. In hindsight, South Dakota could have easily set up a visitors’ registry to provide a valuable service to the understanding of COVID-19 spread.
However, a study conducted by researchers at San Diego State University provides an indirect way to get an accurate estimate of the numbers.
Please do read the referenced paper on the study yourself; below I will summarize the main findings and the methodology used. Predictably, Governor Kristi Noem has called the paper “fiction” and issued a statement that “… this report is nothing short of an attack on those who exercised their personal freedom to attend Sturgis.” The tragic irony is that Sturgis was promoted as “a celebration of American freedom and individual liberties.”
First, the bottomline.
- The study estimates that a total of 266,796 coronavirus cases in the US can be attributed to the Sturgis Rally. This is almost exactly a thousand times the official numbers put out by the various state Departments of Health. (Please note that I am not at all suggesting a conspiracy here—the state DOHs have other concerns besides tracking the cases to the origin; as noted above, they rely exclusively on self-reporting for the official numbers.)
- The estimated number of Sturgis-linked coronavirus cases is 19% of the 1.4 million new cases that the US as a whole experienced between August 2 and Sept 2.
- Assuming conservatively that all the Sturgis-linked cases are non-fatal (even though we already know that this is not true), and assuming a $46,000 statistical cost of a COVID-case (based on an independent estimate), then the Sturgis Rally represents a total cost of $12.2 billion. This is enough to pay each of the 462,182 attendees $26,553.64 not to attend. This health cost of the Sturgis vacation will be paid for by the rest of the country. Our country, not Mexico.
Next, some notes on the methodology of the study.
- Anonymized smartphone data obtained from SafeGraph, Inc., for the period from July 6, 2020 to August 30, 2020 which comfortably straddles the Sturgis Rally.
- The cellphone data enabled the researchers to measure the residence of individuals—defined as a 153 X 153 sq.m area in a night-time location that receives the largest number of GPS pings between 6PM and 7AM—and plot a rough trajectory of the jurisdictions to which they traveled.
- The data was thus used to (1) Measure the number of non-resident visitors to Sturgis during the Rally, (2) Trace the visitors back to their homes, and (3) Measure stay at home behavior of Meade County residents.
- Using this data, it was seen that 90.7% of the attendees came from outside of South Dakota; 18.6% came from bordering states, and the remaining 72.1% from the rest of the country.
- Foot traffic data was also available from SafeGraph. This data was used to locate individuals within dining establishments, grocery stores, liquor stores, gas stations, etc. In addition to tracking behavior of attendees, the data was also used to measure and analyze patterns of stay-at-home behavior of locals as well as social distancing.
- County-level COVID-19 cases of the attendees in their home counties and states was obtained from CDC, Kaiser Family Foundation, NY Times, and Johns Hopkins University.
- Trends in COVID-19 cases for counties (including Meade) that contributed to the largest inflow of attendees to Sturgis are analyzed and presented in the study. As a general rule, higher rates of COVID-19 cases can be seen in the high inflow counties in the period after the rally.
- Various counties that did not contribute significantly or at all to Sturgis were used to control for other factors besides the Sturgis Rally during the study period.
- One significant finding is that the local population of Sturgis and Meade County mingled freely with the attendees during the Rally and the analysis of behavior provides strong evidence that the subsequent substantial spread was due to Sturgis. (Some 625 of Sturgis’ 7000 residents were tested in the week following the rally—of these, 26 tested positive; this extrapolates crudely to 290 who may have been directly affected in Sturgis alone; this number is likely an underestimate, given that the tested group includes all front-line workers who were probably the ones most likely to take precautions.) This suggests that the local population may be at risk for COVID-19 spread unless mitigation strategies are undertaken.
- There is also strong evidence that the Sturgis Rally increased the number of cases in Meade County from 1.38 to 1.54 per 1,000.
- Using a synthetic control method, the study estimates that the entire state of South Dakota has a rise in COVID-19 cases from 1.10 to 1.24 per 1,000 in the week following the Sturgis Rally. These effects are statistically distinguishable from background factors.
- Turning to the nation itself, the study uses a “dose-response-difference-in-differences model” to assess whether counties that sent a relatively high number of Sturgis attendees saw increases in community spread following the event. For example, one finding is that highest relative inflow counties from Arizona, California, Colorado, Iowa, Minnesota, Nebraska, Nevada, Washington, and Wyoming saw a 10.7% increase in COVID-19 cases more than three weeks following the opening of the Sturgis Rally.
- Moderate inflow counties from a diverse set of counties experienced a 7% increase in COVID-19 cases. Again, these effects are statistically significantly different from the low-to-none inflow counties.
A final note. Please remember that correlation is not causation even when statistically significant. The paper of the study has not yet been peer-reviewed. I assume (hope) that the community of sociologists, epidemiologists, statisticians, and others will review the paper carefully and provide apolitical critique and more careful analysis. I hope also that this study will motivate other independent methods to corroborate or refute this study. All said and done, the very existence of a few of us may depend on analyzing the data, and planning and implementing policies to mitigate further spread.