Monday, October 26, 2020

Characteristics of High Risk COVID-19 Settings (Superspreader Events)

With respect to COVID-19 (SARS-CoV-2, or coronavirus), 0ne of the major factors driving the pandemic are superspreader events, or SSEs, where one gathering of people results in numerous infections (which can then go on and spread the chain of transmission to other people, etc). With events like this, one infectious individual can lead to many, many positive cases. Estimates suggest that ~10% of infected individuals can lead to ~80% of additional transmissions. This outsized distribution isn't unique to disease; for example, 1% of individuals in the US have ~40% of the wealth, and in general this phenomenon is known as the Pareto Principle.

With this in mind, avoiding SSEs is a major step towards ending the pandemic. Koen Swinkels has created/maintained a database on SSEs, logging information regarding SSEs around the world.

My goal here isn't any fancy work or statistically savvy analysis. I just want to provide some slicing of data on what types of events, or characteristics of gatherings, are the highest risk for spreading an outbreak of COVID-19. In other words, what things should you be avoiding as to not catch or spread COVID? As a high risk individual myself (I have psoriatic arthritis, and the medication I take for it makes me immunocompromised), assessing this risk is especially important to me.

Characteristics of SSEs

The data has some descriptive information, like whether the event was inside or outside, during the country's flu season, etc. So looking at that first:

# of Incidents# of CasesCases/Incident
RankIndoor / OutdoorCount SSE% CountRankIndoor / Outdoor# Cases% CasesRankIndoor / OutdoorCases/Incident
1Indoor1,46495.7%1Indoor177,49291.7%1Indoor / Outdoor253
2Indoor / Outdoor634.1%2Indoor / Outdoor15,9388.2%2Indoor121
3Outdoor30.2%3Outdoor390.02%3Outdoor13

This is something that has been demonstrated since April - outdoor events are wayyyyy safer than indoors, no matter how you look at it.

# of Incidents# of CasesCases/Incident
RankFlu SeasonCount SSE% CountRankFlu Season# Cases% CasesRankFlu SeasonCases/Incident
1Yes1,13174.1%1Yes163,42784.6%1Yes144
2No30520.0%2No22,86011.8%2Borderline76
3Borderline905.9%3Borderline6,8013.5%3No75

There does appear to be a high overlap with traditional flu season - which has led to fears for months of a 'twindemic' with both COVID and the flu circulating at once. The good news is that the flu has almost disappeared - so the measures put in place to stop coronavirus spread have potentially limited the flu, which points to the factors that lead to season flu spikes (colder weather, more indoor gatherings) exacerbating COVID spread, even without the flu itself spreading.

# of Incidents# of CasesCases/Incident
MappingCount SSE% CountMapping# Cases% CasesRankMappingCases/IncidentMultiple of "Neither"
Vocalizing?21012.4%Vocalizing?44,25619.5%1Both?2121.92
Refrigeration?1146.7%Refrigeration?23,24610.2%2Vocalizing?2111.90
Both?774.5%Both?16,3627.2%3Refrigeration?2041.84
Neither?1,29276.3%Neither?143,02663.0%4Neither?1111.00

Vocalizing (singing, shouting, etc) and refrigeration (colder conditions) have a definitive impact - having one or both of these characteristics meant that the event had an almost 2x increase on how many people got sick.

Unfortunately determining usage of masks, social distancing, ventilation, etc is prohibitively difficult via the sourced news articles, so I can't answer those questions through SSEs. 

Settings of Events

First, I mapped the events to more general settings - whether it was at home, at work, in a healthcare or religious setting, etc.

# of Incidents# of CasesCases/Incident
RankMappingCount SSE% CountRankMapping# Cases% CasesRankMappingCases/Incident
1Living Space (Shared)57437.9%1Living Space (Shared)84,55843.6%1Gathering (Large)612
2Healthcare43528.7%2Healthcare43,11422.2%2Religious415
3Workplace19012.5%3Workplace27,64314.3%3Ship256
4Gathering (Small)573.8%4Religious13,6907.1%4Living Space (Shared)147
5Public Interaction442.9%5Gathering (Large)11,6276.0%5Workplace145
6Living Space (Home)422.8%6Ship4,0902.1%6Healthcare99
7Bar/Restaurant372.4%7Bar/Restaurant2,1211.1%7Funeral/Wedding58
8Religious332.2%8Gathering (Small)1,8060.9%8Bar/Restaurant57
9School/Childcare302.0%9School/Childcare1,5040.8%9School/Childcare50
10Funeral/Wedding201.3%10Funeral/Wedding1,1510.6%10Other40
11Gathering (Large)191.3%11Living Space (Home)1,0300.5%11Travel33
12Ship161.1%12Public Interaction8730.5%12Gathering (Small)32
13Other100.7%13Other4030.2%13Living Space (Home)25
14Travel80.5%14Travel2640.1%14Public Interaction20

The vast majority of cases occurred in a shared living space (shelters, jails, nursing homes, dorms, or military base, more on this later), healthcare setting with a prolonged stay (rehab facility, medical or mental health facility, or hospital), or workplace (wide variety, more on this later). However, the highest outbreaks per event were large gatherings (conferences, concerts, sporting events), religious events, or being on a ship.

Breaking down non-optional exposure settings, housing and workplaces:

# of Incidents# of CasesCases/Incident
HousingCount SSE% CountHousing# Cases% CasesHousingCases/Incident
Nursing Home34155.8%Prison50,05859.7%Prison253
Prison19832.4%Nursing Home30,53836.4%College94
Household315.1%Housing2,1282.5%Nursing Home90
Housing294.7%Housing (Shelter)7570.9%Housing (Shelter)76
Housing (Shelter)101.6%Household2150.3%Housing73
College20.3%College1880.2%Household7

As is known by now, explosive tragic outbreaks are extremely common in both nursing homes and prisons.

# of Incidents# of CasesCases/Incident
WorkplaceCount SSE% CountWorkplace# Cases% CasesWorkplaceCases/Incident
Food Manufacturer11460.0%Food Manufacturer21,83379.0%Food Manufacturer192
Workplace (Non-Office)3317.4%Workplace (Non-Office)2,5199.1%Factory109
Workplace (Office)3116.3%Workplace (Office)1,9857.2%Workplace (Non-Office)76
Factory126.3%Factory1,3064.7%Workplace (Office)64

Additionally, offices have the lowest incidence rate - food manufacturing plants and factories have much higher risk, with non-office work environments also having worse outbreaks than office settings.

Now breaking down optional exposure settings, types of voluntary gatherings:

# of Incidents# of CasesCases/Incident
MappingCount SSE% CountMapping# Cases% CasesMappingCases/Incident
Gathering (Small)5759.4%Gathering (Large)11,62779.7%Gathering (Large)612
Funeral/Wedding2020.8%Gathering (Small)1,80612.4%Funeral/Wedding58
Gathering (Large)1919.8%Funeral/Wedding1,1517.9%Gathering (Small)32

While there were far more instances of small gatherings (parties, meals, community center events), the smaller number of participants greatly lowered the rate of increased cases - large gatherings (conferences, concerts, sporting events) have led to explosive growth, with an average of over 600 cases per SSE.

A more granular view of all event types:

# of Incidents# of CasesCases/Incident
RankMappingCount SSE% CountRankMapping# Cases% CasesRankMappingCases/Incident
1Healthcare Facility40626.8%1Prison50,05825.8%1Live Events1,831
2Nursing Home34122.5%2Healthcare Facility39,95720.6%2Market714
3Prison19813.1%3Nursing Home30,53815.8%3Demonstration500
4Food Manufacturer1147.5%4Food Manufacturer21,83311.3%4Religious415
5Small Gathering392.6%5Religious13,6907.1%5Military341
6Religious332.2%6Live Events7,3223.8%6Ship256
6Workplace (Non-Office)332.2%7Ship4,0902.1%7Prison253
8Workplace (Office)312.0%8Hospital3,1571.6%8Food Manufacturer192
8Household312.0%9Market2,8541.5%9Farm174
10Hospital291.9%10Workplace (Non-Office)2,5191.3%10Hospital109
10Housing291.9%11Housing2,1281.1%11Factory109
12Bar/Nightclub271.8%12Workplace (Office)1,9851.0%12Summer Camp100
13School221.5%13Bar/Nightclub1,9371.0%13Healthcare Facility98
14Sports: Participation201.3%14Military1,7040.9%14College94
15Ship161.1%15Factory1,3060.7%15Nursing Home90
16Wedding140.9%16School1,1480.6%16Workplace (Non-Office)76
16Shopping140.9%17Small Gathering1,0890.6%17Housing (Shelter)76
18Factory120.8%18Demonstration1,0000.5%18Housing73
19Choir110.7%19Wedding7980.4%19Bar/Nightclub72
20Housing (Shelter)100.7%20Housing (Shelter)7570.4%20Workplace (Office)64
20Restaurant100.7%21Choir4300.2%21Funeral59
22Unknown90.6%22Shopping3790.2%22Wedding57
23Conference80.5%23Conference3740.2%23School52
23Travel80.5%24Funeral3530.2%24Conference47
25Funeral60.4%25Summer Camp3000.2%25Community Gathering46
25Community Gathering60.4%26Community Gathering2760.1%26Hospitality44
27Military50.3%27Travel2640.1%27Choir39
27Gym50.3%28Sports: Participation2530.1%28Court38
27Childcare50.3%29Unknown2290.1%29Gym34
30Live Events40.3%30Household2150.1%30Travel33
30Market40.3%31College1880.1%31Small Gathering28
32Summer Camp30.2%32Restaurant1840.1%32Shopping27
32Hotel30.2%33Farm1740.1%33Unknown25
34Demonstration20.1%34Gym1710.1%34Restaurant18
34College20.1%35Hospitality880.05%35Sports: Participation13
34Hospitality20.1%36Childcare560.03%36Childcare11
37Farm10.1%37Court380.02%37Salon9
37Court10.1%38Hotel230.01%38Hotel8
37Salon10.1%39Salon90.005%39Household7

Indoor, full live events (audiences for concerts or sports, with no masks or distancing) are clearly huge risks (1,831 cases per instance) - contrasted against participating in sports as not as high risk (13 cases per instance).

As we already knew, prisons, nursing homes, and healthcare facilities (like rehab centers, or long-term care) are tragically huge common hotspots, and make up 62.2% of all cases tied to SSEs.

Funerals and weddings don't appear as often as you might think given high profile news coverage (1.3% combined); same goes for demonstrations (only 2 instances, or 0.1%). Meanwhile, smaller, more socially optional gatherings like going to bars/clubs, restaurants, religious services, and small private gatherings make up a larger share of events (7.3%).

Even the granular look at SSEs shows some slight Pareto tendencies: 72% of events make up 81% of cases associated with those events.

Takeaways

  • Outside > inside
  • Smaller groups > larger groups
  • Limit shouting/singing, and warmer environments are better
  • Be more vigilant during your area's traditional flu season
  • If possible, avoid shared living quarters

Citation

Last updated on 10/10/20, I pulled the data down on 10/23/20 (my edits linked here for transparency)

Swinkels, K. (2020). COVID-19 Superspreading Events Around the World [Google Sheet]. Retrieved from https://docs.google.com/spreadsheets/d/1c9jwMyT1lw2P0d6SDTno6nHLGMtpheO9xJyGHgdBoco/edit?usp=sharing