Tables Showing the Changes in Breakthrough Events

We split this by event so you can more easily see the changes.  As Dave notes, there is clearly something wrong with DOH’s revisions to the death numbers, so expect further changes eventually.  And note that they cleaned up the two spikes in unvaxed deaths as well.  As I noted in the original tweet, the effect of the revisions is to make the vaccines and boosters look less effective at preventing infection and hospitalizations.  Just picking one week, for example, July 17, the percent of cases in the vaxed and vaxed and boosted groups goes from 70% to 73% and the proportion of hospitalizations from 64% to 67%.

Dave’s notes:

• On 10/06/2022 the Minnesota Department of Health (MDH) revised their reporting on Covid breakthrough cases, hospital admissions, and deaths, with the changes extending back to the start of breakthrough reporting on 5/02/2021. The net effect of the changes in reporting were to shift cases and hospital admissions from the unvaccinated to the vaccinated, and to the boosted. Surprisingly, reported deaths were little changed, and in fact shifted a little towards increased unvaccinated deaths.
• Covid breakthrough data is reported on the MDH Situation Update for COVID-19 web page, available here: https://www.health.state.mn.us/diseases/coronavirus/stats/vbt.html. MDH added a note on 10/06/2022 explaining the revisions, stating “As of October 6, an improved method to match patient data and vaccination records was implemented. This methodology was applied to all cases going back to the start of reporting on vaccine breakthrough cases.”
• Fig. 1: Comparison of reported Covid cases on 9/29/2022 and 10/06/2022. The first set of columns highlighted in green show the unvaccinated cases reported for each week on 9/29/2022, and the revised unvaccinated cases as reported on 10/06/2022, along with the changes in cases, and as a proportion of the cases reported on 9/29/2022. The vaccinated but not boosted cases are highlighted in blue, and the vaccinated and boosted cases highlighted in red. Note that in most weeks cases are moved almost exactly from the unvaccinated to the vaccinated, and to the boosted. The magnitude the changes is highest on 1/09/2022, the peak of the Omicron wave. There are also very large revisions the week of 6/12/2022 and 8/07/2022, where we have previously noted what appeared to be anomalously high levels of unvaccinated cases.
• Fig. 2: Comparison of reported Covid hospital admissions on 9/29/2022 and 10/06/2022. In this table we display the changes in reported hospital admissions in a format identical to that used for cases in Fig. 1. Note that for admissions the pattern of revisions is similar to that of cases; admissions are shifted from the unvaccinated to the vaccinated and to the boosted almost exactly. We see the same large revisions on 1/09/2022, 6/12/2022, and 8/07/2022 as we had for cases, but also a large change on 3/06/2022.
• Fig. 3: Comparison of reported Covid deaths on 9/29/2022 and 10/06/2022. In contrast to the pattern of changes seen for cases and hospital admissions, the changes in reported deaths largely consists of shifting vaccinated deaths to the unvaccinated and to the boosted. It is hard to imagine data base errors which would incorrectly flag a Covid death as vaccinated when they were actually unvaccinated. Missing a booster shot record is perhaps more understandable. These changes to death records really do not make much sense.
• We have submitted an inquiry to MDH asking for more details about the revisions, and which populations are affected. For example, we have previously been informed that MDH does not have access to the vaccination records of many states. Minnesotans who were vaccinated in Florida or Arizona, for example, will not appear in the data as vaccinated unless they were interviewed by an MDH case tracker. MDH also does not have access to Veterans Administration vaccination records, or of tribal health authorities. At least, that is what we were told in the past. There were also very minor revisions to the size of the unvaccinated, vaccinated, and boosted populations on 10/06/2022. However, the changes were so small we do not believe that they represent the addition of these missing state, VA, or tribal records. Rather, they appear to be just minor revisions to the previously used vaccination records.