CO₂, Infection Risk, and the Worst Air I’ve Ever Measured
Published:
I brought an air quality monitor to a Rhode Island State House committee hearing and watched the CO₂ hit 6,000 ppm — the worst air quality I have ever personally measured. Here’s what that means for infection risk, and the model I built to show it.
–
This past April, I brought an air quality monitor to the Rhode Island State House to testify, in my personal capacity, about a bill mandating indoor air quality inspections in public schools. Room 35 was packed — well over 100 people in a space designed for maybe 70. I left my monitor in the corner and watched from the hallway as CO₂ climbed past 2,000 ppm, past 5,000 ppm (OSHA’s permissible exposure limit for industrial settings), and settled at 6,000 ppm for the rest of the night. It was the worst air quality I have personally measured.
That experience prompted me to build the model below. Using the Wells-Riley equation — the foundational framework in airborne transmission epidemiology — it estimates expected infections from a single infectious person in a shared room, across the full range of indoor CO₂ concentrations. CO₂ is one of the best real-time proxies we have for ventilation quality: the same physics that allows exhaled CO₂ to accumulate also allows exhaled aerosols to accumulate. The model uses quanta emission rates from the published literature for both COVID-19 and influenza, assumes a realistic activity mix (mostly quiet, with brief talking), and includes confidence intervals to reflect genuine uncertainty across strains and individuals.
The takeaway from that hearing room: at 6,000 ppm, with 60 people and one infectious person over four hours, the math suggests roughly 9 to 10 people would have gone home infected. Not a worst case — just the arithmetic of shared air. The tool below lets you see how that number changes across the full range of ventilation conditions, and where the thresholds that matter actually sit.
Model inputs and assumptions
The model assumes a room of 60 occupants (1 infectious, 59 susceptible) sharing air for four hours, with each person talking for approximately 2.5 minutes and seated quietly otherwise. Breathing rate is set at 0.30 m³/hr, consistent with light sedentary activity. Ventilation is derived directly from CO₂ concentration: each person produces approximately 18 litres of CO₂ per hour, and the steady-state excess above outdoor background (420 ppm) is used to back-calculate total room airflow. Quanta emission rates — the currency of the Wells-Riley model — are time-weighted averages of resting and talking emission, drawn from Buonanno et al. (2020) and Miller et al. (2021): 28.4 quanta/hr for COVID-19 and 13.6 quanta/hr for influenza at the central estimate. Confidence intervals reflect a ×0.5 to ×2 range on the central quanta value, capturing uncertainty across pathogen strains, individual variation, and measurement methodology. The model does not account for masking, room geometry, near-field exposure, or variation in susceptibility.