Where a Minnesotan Got Coronavirus

While I wait for the case by date of specimen collection data to season (i.e., catch up to reporting lags) I thought it might be interesting to look at source of exposure.  The Minnesota weekly reports have included a page showing this for some time.  I looked the reports from July 2 to August 20.  I was curious to see if the percent of cases deriving from settings where you might expect masks to make the most difference has changed in recent weeks, when the mask mandate is supposedly doing its magic.  Interestingly, Minnesota does not break out one of the most common sources of transmission–the home.  Apparently that is included in the “community” buckets.  Workplaces, schools and retail settings also aren’t broken out, again surprising, nor is the infamous and dreaded restaurant and bar sector, where the virus lurks like a bad pickup artist.  So not the best set of buckets, although just this past week, we finally got definitions for the categories, but not sub-bucket percents.

What they do have is congregate care, where you think by July people were wearing masks pretty strictly; health care workers, who presumably wear masks and know how to wear them; and community outbreak, community known contact and community unknown contact.  Community known and unknown contact are default categories, if a case doesn’t fit elsewhere it goes in one of those buckets.  Outbreak we now learn does include bars, restaurants, gyms and sports, workplaces and presumably stores.  Again, since homes may be in the community bucket, presumably known contact, it is kind of odd that they don’t break that out separately.  But in any event, let’s see what changes there have been, especially in community outbreak.

These are cumulative percents, so a change from week to week in a cumulative number may reflect a bigger change than it seems like.  Community outbreak started at 13% on July 2, stayed there for three weeks, dropped to 12% on July 24, to 12% a week later, to 10% a week later, to 9% the last two weeks.  So if the Health Department is right that it would take three weeks or more to see the effect of masks, community outbreak as a percent had already begun dropping.  But of course, anytime you start looking at data from our state, you are going to see bizarre things that you don’t expect.  I have no idea why, but from July 24 to July 31, the number of cumulative cases in the community outbreak bucket actually declined by 68 cases.  How does that happen?  I suspect some definitional change, some contact tracing outcome change, I simply don’t know, but those 68 cases probably would need to go out in the earlier reports as well.  Makes it impossible to actually see a trend.

And it doesn’t stop there, in the August 20 report cumulative infections among health care staff miraculously declined from 4735 the week before to 2379.  Why?  Some but not all appears to be because with the definition explanation there was also a bucket assignment change–it appears that maybe now congregate care includes staff and residents, so some of those health care worker cases got moved to the congregate care bucket.  The other big jump in that week was in the community known contact group, so it would seem that many health care worker cases got moved to the bucket.

The real point is, once again, pretty impossible to do a trend analysis when definitions change, the composition of buckets changes, and cases get moved around without telling us how many moved or when.  So once again, late at night, I am frustrated by inability to get data that would help all of us understand important trends.  And it made it pretty impossible for me to see if the data would help identify any effect of masks.  The total percent of community transmission in all three buckets has risen since the mask mandate, but with the definitional issues, not sure that tells us anything.