Breakthrough Event Timing

So we think we have learned something new from DOH about how breakthrough events are presented.  According to an answer received by Dave Dixon, the events are all reported by date of specimen collection.  What this means is that we don’t have actual date of hospitalization or death (in the typical file released, we did get one file, which apparently we weren’t supposed to get, that did have the dates of those breakthrough events).  But if we are understanding DOH accurately, what that means is that for any week for which data is reported, you should have an accurate case rate.

Getting this answer solves another mystery, which is why the data for prior weeks was often adjusted in the current week’s data.  These adjustments would be inconsistent with a date of report method.  You can see the significance of these changes for yourself by, for example, looking at our table of breakthrough events from the last three weeks.  Just pick out a week on the table, say September 19.  In the most recent release, 5015 breakthrough cases were assigned to this week; the week before it was 4928 cases and the week two weeks before also was 4928.  So both the number and extent of changes is variable, which I suspect reflects how much effort the state puts into identifying breakthrus and I doubt it is a priority, as it is contrary to the messaging.  In general, the changes over time should be adding to the breakthrough columns and subtracting from the non-breakthroughs, since much of this is just bucket shifting as someone is identified as fully vaxed.

In any event, assuming that we understand DOH correctly, in the most recent week’s table, calculating case rates gives the following, and keep in mind that this is not by age group, this is across all cases.  So November 14 is the most recent week of data, the case hospitalization rate fro the vaxed is 3.1%, for the unvaxed is 6%.  The case death rate is .54% for the vaxed, .84% for the unvaxed.  But now let’s go back a few weeks, where the data may be more complete.  For the week of October 17, the vaxed hosp rate is 4.1%, while that for the unvaxed is 5.8%.  The vaxed death rate is 1.2% versus 1.04% for the unvaxed.  Keep in mind the age structure issue, but also that we don’t know the relative number of undetected infections in each group.  Vaxed people could have more undetected infections.  And we don’t know the overlap of prior infection and vaccination status, which likely also is different between groups.

Dave is working further on this, so more to come.