If initial viral load makes a difference one would expect to see shorter time from infection to diagnosis/hospitalisation in cases which are transmitted within households. There is suggestive evidence in this paper which includes data on the serial time for household (4.03 [3.12, 4.94]) and non-household (4.56 [3.85, 5.27]) secondary infections. The number in square brackets are the 95% CI.
This is fairly weak evidence that there is a difference and also gives some weak indication as to what the maximum effect of initial viral load might be.
The raw data from this paper, for example, might be used to give more information on this and also severity which is more what we’re interested in—the Tianjin data appears to be fairly complete albeit with only 135 cases.
It isn’t clear—that’s a good point and would suggest that the upper bound might actually be higher than it appears at first glance. If we take 10% of infections being hospital based (which might not be accurate as that statistic is from South Korea and the above paper is in China outside Hubei) then 16% of the outside-the-home transmission might be hospital based.
I should say that only 284 of the 468 transmission events are included in either household and non-household. I don’t know what the other 40% of cases were but I guess the researchers weren’t able to identify the relationship from the public data that they were using. It does appear that this undefined 40% has a lower serial interval than either of the two defined groupings as the serial interval of all cases together is lower 3.96 [3.53, 4.39].
If initial viral load makes a difference one would expect to see shorter time from infection to diagnosis/hospitalisation in cases which are transmitted within households. There is suggestive evidence in this paper which includes data on the serial time for household (4.03 [3.12, 4.94]) and non-household (4.56 [3.85, 5.27]) secondary infections. The number in square brackets are the 95% CI.
This is fairly weak evidence that there is a difference and also gives some weak indication as to what the maximum effect of initial viral load might be.
The raw data from this paper, for example, might be used to give more information on this and also severity which is more what we’re interested in—the Tianjin data appears to be fairly complete albeit with only 135 cases.
EDIT: added link to 2nd paper
Do you know if outside-the-home includes hospital transmission? That could skew things severely.
It isn’t clear—that’s a good point and would suggest that the upper bound might actually be higher than it appears at first glance. If we take 10% of infections being hospital based (which might not be accurate as that statistic is from South Korea and the above paper is in China outside Hubei) then 16% of the outside-the-home transmission might be hospital based.
I should say that only 284 of the 468 transmission events are included in either household and non-household. I don’t know what the other 40% of cases were but I guess the researchers weren’t able to identify the relationship from the public data that they were using. It does appear that this undefined 40% has a lower serial interval than either of the two defined groupings as the serial interval of all cases together is lower 3.96 [3.53, 4.39].