Our conclusions rest on some key assumptions and could change if these assumptions are incorrect.
We assumed that viral load is a proxy for infectiousness. We tested different assumptions that related viral load to infectiousness, so that we could ensure our results did not depend on one particular relationship. We erred toward estimating that infectiousness lasted longer after-symptom improvement to best quantify the impact of potentially reduced isolation times on onward transmission potential. Should we include additional data on infectiousness—such as viral culture measurements that determine whether an individual is actively transmitting live virus—indicating the duration of infectiousness is reduced, then both the updated and previous guidance would be more protective than we estimated.
Our results are sensitive to testing practices. If, under the previous guidance, people only test one time when their symptoms start, rather than taking two tests two days apart as we assumed, then the updated guidance might prevent approximately 9 more percentage points of overall transmission potential than the previous guidance. This is because many people with COVID-19 get a false negative result if they only take one antigen test immediately after symptoms start, and thus do not take post-isolation precautions.
We used data on which symptoms study participants had each day to infer when symptom improvement occurred, but study participants were not asked directly whether they felt symptoms were improving. If people with COVID-19 feel better sooner than we estimated, then there would be a larger difference between the updated and previous guidance; this generally would make the updated guidance less protective than we estimated.
Two distinctions between the updated and previous guidance are not captured in our analysis. The updated guidance calls for isolation to continue until symptoms have been improving for 24 hours; the previous guidance referred to whether “symptoms are improving” without specifying any minimum duration of improvement. If captured in the analysis, these differences could make the updated guidance more protective than we estimated.
Additionally, the previous guidance gave more specific recommendations than the updated guidance about what actions to take in the post-isolation period. This may make the updated guidance less protective than we estimated.
Finally, our analysis focused on modeling the consequences of the updated guidance versus the previous guidance, assuming people are equally likely to follow the updated versus previous guidance and that isolation and post-isolation precautions are equally effective under both sets of guidance. We did not model changes in the extent to which people follow COVID-19 isolation and post-isolation guidance. We also did not compare the updated and previous guidance for respiratory diseases other than COVID-19.
Few study participants included in this analysis developed symptoms severe enough to seek medical care. The transmission potential of hospitalized COVID-19 patients is outside the scope of this analysis. The updated guidance applies only to community settings, not to other settings such as healthcare settings.
