Skip to main content

Cohort Outcome Analysis

By August 24, 2022Commentary

You may recall that near the start of the epidemic, even before Dave came onboard with his superior data and analytic skills, we tried to show what the likely outcomes within a group of cases were.  What has been frustrating is that the state has the exact data to show outcomes by case but won’t release it.  So we have to infer rates of outcomes for a group of cases by average dates from positive test to hospitalization and/or death. Dave has added subgroups here for breakthroughs and reinfections.  But you will recall that there is overlap, probably very large overlap, in those two groups, which again, DOH could easily identify for us, but refuses too.  So the overall analysis and the subgroup ones probably have large error bands.

Also as Dave notes, the number and ratio of undetected cases which don’t show up in the official numbers is quite large at this point, but likely has varied during the epidemic, so the rates we are publishing are actually probably far lower.

Dave’s notes:

  1. These charts display the proportion of cases on any given date who go on to be hospitalized or die. We last published this type of cohort analysis on 1/28/2022 here: https://healthy-skeptic.com/2022/02/01/cohort-analysis-january-28/. Since then we have been able to obtain data from the Minnesota Department of Health (MDH) allowing us to also calculate the hospitalization and death rates for breakthrough and reinfection cases. However, there are several important caveats with the data, as noted below.
  2. The charts display the case hospitalization and death rates for all Covid cases (blue curves), which include all breakthrough and reinfection cases.  The reinfection case rates (orange curves) may or may not be vaccinated. MDH does not give us enough data to know which reinfection cases are also vaccinated. Similarly, the breakthrough case rates (green curves) also may include reinfection cases that occurred in vaccinated people. Any reinfections that occur in in vaccinated people will end up being part of the populations for all three categories of cases. We have submitted Minnesota Government Data Practices Act requests in an attempt to get additional data to allow us to eliminate this overlap in cases.
  3. Note also that these case hospitalization rates and case death rates are simply the number of hospitalizations or deaths divided by the number of cases. Only cases which are identified and logged with MDH will show up in the data. Covid cases which are mild enough that the infected person does not get tested, or only takes home tests, tend to increase these rates.
  4. Data for the Total Case Rate curves is obtained from the Minnesota Department of Health (MDH) Situation Update for COVID-19 web page, https://www.health.state.mn.us/diseases/coronavirus/situation.html. Cases by specimen collection date are in the data file “c7day.csv”, available under the Cases Over Time graphic. Hospital admissions by admissions date in the data file “h7day.csv”, available under the Hospitalizations Over Time graphic. Deaths by date of death are in the data file “d7day.csv”, available under the Deaths Over Time graphic. Since we do not have data that follows the outcome of each individual who tests positive, we are forced to compare hospital admissions over a period time to cases over a different period of time, assuming that hospital admissions occurs on average 3 days after the specimen collection date of the positive case. Deaths are similarly analyzed assuming that death occurs 14 days on average after the specimen collection date of the positive case.
  5. Data for the Reinfection Case Rates and Breakthrough Case Rates was obtained through a Minnesota Government Data Practices Act request. The data files were received on 7/28/2022, and have date stamps indicating they include data in MDH’s possession as of 7/26/2022. Data after approximately 7/01/2022 is likely to be incomplete. These data files list the reinfection and breakthroughs cases by date of specimen collection, along with Yes/No data flags indicating whether each case was hospitalized or died. Because we know the outcome of each positive case we do not have to assume lags from specimen collection date to hospitalization or death, as we have to for the total case rate curves.
  6. Case hospital admission rates are displayed in Fig. 1, 2, and 3. Fig.1 displays the entire pandemic, with the high hospital admission rate above 40% early in the pandemic very likely due to low testing rates. The hospital admission rates around 5% throughout 2020 are more representative. It is unclear why rates spiked to 15% in June, 2021. This may be due to lower testing rates as well. The first reinfection case was identified on 7/03/2020, and the reinfection case hospitalization rates are substantially higher than the total case hospitalization rate for many months. It is possible that people tended not to get tested for a reinfection case, causing the reinfection hospitalization rate to be higher than for the total cases. The first breakthrough case was identified on 2/01/2021, and also had higher hospitalization rates than total cases for several months, although lower than the rates for reinfections. The spike in the reinfection hospitalization rate at the far right end of the curve is likely due to incomplete data reporting by MDH.
  7. Fig. 2 displays the case hospital admission rates for specimen collection dates during the Delta variant wave in cases, from 71/1/2021 to 11/30/2021. During this time period the hospital admissions rates stabilized in the range of 4% as cases rose, for all three categories.
  8. Fig. 3 displays the case hospital admission rates for specimen collection dates during the Omicron variant wave in cases, from 12/1/2021 to 3/18/2022, the date by which cases receded to a low level. We can see that during this time period the case hospitalization rate fell by half, from 4% to less than 2%, before rising in late winter as the number of cases fell.
  9. The charts for the case death rates are shown in Fig. 4 through Fig. 6, using similar time periods as used for case hospitalization rate charts. Fig. 4 display the case death rates for the entire pandemic, the percentage of positive cases who are reported to die of Covid. Just as for case hospitalization rates, the high rates in early 2020 are likely due to low testing rates, as well as for the initial reinfection and breakthrough cases. As testing increases the case death rate dropped to between 1% and 3% throughout most of 2020 for total cases, while reinfection and breakthrough cases saw higher death rates. In 2021 the death rates for all three categories was typically between 1% and 2%, and falling below 1% for 2022.
  10. Fig. 5 displays the case date rates for specimen collection dates during the Delta variant wave in cases, from 7/1/2021 to 11/30/2021. During this time period death rates for all three categories are fairly similar, although the relatively few number of reinfection deaths cases many data points to be zero.
  11. Fig. 6 displays the case death rate for specimen collection dates during the Omicron variant wave in cases, from 12/1/2021 to 3/18/2022. We can see that the case death rate fell to less than 0.3% toward the end of the time period for all three categories before rising in March 2022 as the number of cases fell.

Leave a comment