Please download the excel sheet here:
Version July 10.
Summary, Implications, and Recommendations
INTRODUCTION
This calculator determines the number of new COVID-19 infections in a space (classroom, etc.) due to aerosol transmission, accounting for infectious aerosol in the breath of asymptomatic individuals. The aerosols, at sizes 7 microns and below, are emitted from the normal action of breathing and talking -- they have lifetimes of hours in air, and widely disperse throughout indoor spaces. Symptomatic individuals emit these aerosols too, but we assume symptomatic individuals self-quarantine and are not in our classroom spaces. The model accounts for occupancy reduction, heating-ventilation-air conditioning (HVAC) characteristics, and mask usage.
This tool combines work on ventilation done by Charles Stanier and Tom Peters for the ventilation sub-committee at the University of Iowa, with new quantitative infection probability tools released by Jose Jimenez, Shelly Miller, and others [Jimenez, Miller Airborne Transmission Estimator (https://tinyurl.com/covid-estimator)]. The Jimenez calculator requires a number of inputs, which may be difficult for decision makers to evaluate; therefore, we have put University of Iowa specific values in place, in line with the discussions and tests of the ventilation committee over May and June 2020.
The calculator takes as input, the ventilation type of the room, the occupant density, mask compliance, and the number of contagious people in the room. It takes into account transmission from asymptomatic cases only, and assumes 100% compliance of self-quarantine when symptoms appear. The results from this calculator agrees to those from the Jimenez estimator within expected variation. The calculator has also been used to reproduce the Skagit Valley choir outbreak (this calculator underpredicts the Skagit valley case, and thus may be a conservative tool).
RESULTS & IMPLICATIONS
Under realistic scenarios, with 80% compliance with masks, the number of new infections on campus from aerosol transmission inside classrooms buildings could range from 10 to 31 cases per day (650 to 2,015 for a 13 week semester). These infection rates (10-31 cases per day), because of the high ratio of asymptomatic cases and cases not confirmed by testing, may only yield one to three test-confirmed cases per day. This can be compared to rates of new test-confirmed cases (using July 8 weekly average) in Iowa (343/day) or Johnson county (23/day). At 90% mask compliance, infection rates are reduced by about 9%. Under nonindeal cases (limited ventilation, large amounts of talking, low mask usage, and normal occupancy levels), infection risks can be much higher, exceeding 10% per hour per person.
Calculations are based on the assumption of 0.74% of people coming into classrooms asymptomatic but COVID-19 contagious. This value (0.74% contagious and asymptomatic) is a an estimate for the start-of-semester using an average of Iowa, Illinois, and Johnson County for July 8. The model assumes physical distancing and classroom occupancy reduction by a factor of two. The estimate above further assume that classrooms without mechanical ventilation are not used for holding classes in fall 2020. Worst case infection rates could rise to about 5 times higher if the case load in the campus community increases, driven by the number of infectious people and their associated viral bioaerosol emissions (i.e. reaching 50-150 total cases per day, with 5-14 per day test-confirmed cases).
As a consequence, the University should expect and plan for significant community transmission from the aerosol inhalation route, driven by emissions from asymptomatic people talking. Surveillance, contact tracing and continued test of ventilation rates in spaces should be used to monitor this as the semester progresses. As the situation evolves, the University should be ready to respond with further changes in protective behaviors, classroom scheduling, shift-to-online, occupancy changes, etc.
The type of vocalization in the room is critical. Loud talking is thought to emit over 80x the levels of no talking. Accordingly, some of the recommendations below focus on the most frequent talkers -- our professors and classroom instructors. Another consequence of the important of loud talking is that off campus gatherings can have a significant infection rate from asymptomatic aerosol transmission. Attendance of indoor social gathering (e.g., bars, dance clubs, house parties, etc.), with typical occupant densities and ventilation, and no masks, could generate 342 cases per week (31 confirmed cases per week -- assumptions in Note 10 on input/output sheet). On a weekly basis, this is 2-7 times higher than the case load that the model generates from classroom spaces.
This does not include other transmission modes (contaminated surfaces, droplets from cough, sneeze). And while we have done some comparison cases for off campus infection rates, our focus has mainly been on indoor classroom spaces, in order to assess mitgation strategies and assist in campus planning. Separate work is needed for apartments, bars, social events; the summary values stated above exclude on campus transmission in dorms, dining halls, athletic events, hallways, elevators, restrooms, CAMBUS, and other public spaces.
Although hospitalization and death rates for the 20-30 age group are low, (e.g. 1% hospitalization rate for 20-30 yr old), this large number of cases could be a significant source of community spread. Furthermore, there are non-traditional students, students with preexisting health problems, and staff and faculty to be concerned about.
I have verified these calculations and believe they are free of errors. They have been reviewed by several expert reviewers (see the Q&A tab, and Change Log & Review tab) and by the UI ventilation sub-committee. Generating the responses to questions has raised my confidence in the tool; furthermore, it has not changed the overall tool structure or the underlying formulas. It has resulted in refinement of the inputs and clearer interpretation of the results.
Some recommendations based on these results:
1. keep people who have symptoms at home. Also keep asymptomatic shedders at home, through testing, contact tracing, and education
2. pay most attention to those who are talking. Taking off PPE while speaking (as most public officials seem to do) significantly increases risk.
For example, to address #2, adjusting the current training materials to emphasize that the most important time to wear your mask, is when you are talking.
For example, to address #2, instructors should practice talking with their mask on, and search around to get masks that fit well while they are talking.
3. promote interaction through means other than talking (written work, virtual collaboration, prerecorded lecture segments?).
4. do not schedule class in the rooms with poor ventilation
5. limit total number of hours of in-person class for any student
6. practice active surveillance to determine the extent to which in-classroom transmission is occurring
7. Current University policy seems to allow mask or faceshield in indoor locations. But for the size of aerosols referred to in Miller et al. and related work, masks are better than faceshields. The particles are too small and will be redirected but not trapped by the faceshield. I realize there are pros and cons to faceshields and masks, but for stopping particles smaller than 10 microns during talking and breathing, I don't believe faceshields are effective. A mask requirement rather than either or policy would be a prudent action in light of what we are now learning.
8. Outreach (Johnson County public health, the City of Iowa City, business owners, churches) about amending the "three C's approach" (Closed spaces with poor ventilation, crowded places, close contact settings) to include "loud" settings as something to avoid and/or manage might help us get through the semester with a manageable level of community spread.