That study seems to have an obvious and unacknowledged flaw: they did not control for the growing prevalence of the delta variant this summer even though the time frame of the study exactly coincides with the spread of delta within the united states.

These sorts of analyses are extremely hard to control all confounds for.