The CDC and Minnesota Don’t Make It Easy to Work With Coronavirus Numbers

I wrote a couple of posts last week comparing death rates among different age groups, using Centers for Disease Control data.  It would be useful to be able to make those age group comparisons by reported cases as well.  I would like to be able to do the same thing for my home state as well.  But the CDC has different groups putting out the case and death data, so of course, they don’t use the same age bands.  What seems forever ago now, I wrote a post on understanding all the numbers and rates you may see referred to.  At the end of an epidemic, you would be most interested in population rates and potentially case rates, depending on well you thought you understood the actual number of cases.  And those rates can be helpful as the epidemic proceeds, but should be used with caution, due to uncertainty about the accuracy and completeness of data. For deaths, the CDC uses age bands of under 1, 1 to 4, and then a series of ten year age bands up to 85 and over.  For cases, it is under 18, 18 to 44, 45 to 64, 65 to 74 and 75 and over.  The CDC in the death report gives information on the population in each age bracket so that is helpful, but it is not done for cases report.  So you can kind of do some work to back into deaths as a percent of cases, but it is not easy.  And it is compounded by the fact that the data is reported as of different dates.

So here is my updated analysis, with case data through April 30th and death data through April 28.  In both cases, and more so for deaths, the CDC is dependent on data reported to it from the states, so there is a lag.  On April 30, the CDC said there were a cumulative 812,275 cases.  16,980 were in people under the age of 18, or 2.1%.  286,823 cases were in people aged 18 to 44, or 35% of cases.  283,597 were in the 45 to 64 age group, or another 35%.  88,249 were in the 65-74 age cohort, or 11%, and the 75 and older group had 99,103 cases or 12%.

Now again, remember that there are different data lags.  But let’s try to roughly match up deaths and cases.  For under one year, CDC has 4 deaths, for one to 4, 2 deaths, for 5 to 14, 3 deaths and for 15 to 24 years old, 39 deaths.  This is for the whole country, 48 deaths in the 24 and under age groups, out of a population of 104 million.  On the other end, for people over 75, there have been 20,021 deaths out of a population of around 22 million.   The first place we can try to match up the differing age groups is the 44 and under age bands.  That group represented 303,803 cases, or 37.4%.  It accounted for 958 deaths, 2.8% of all deaths, and a case fatality rate of .31%.  Now let’s go to the other end, the 75 and over age group.  There were 20,021 deaths in this group, out of 99,103 cases.  That is a case fatality rate of 20.2%.  Now I again want to remind you that the true number of infections is likely ten times or greater than reported, or more, especially in the younger cohorts, where cases tend to be asymptomatic and mild.  The ratio is probably not as great for the older cohort.  Again, remember this data is lagged differentially so it isn’t going to be the real calculation, but it gives you a directional sense.

Minnesota tends to use age ranges and median age for reporting on deaths and just gives you a bar graph of percent of cases by age.  So it is very hard to understand what is actually happening.  The state has been very forthcoming about the number of deaths in long-term care facilities, which currently stands at 271 out of 343 total deaths.  Those deaths span an age range, because some people are in those facilities for long-term rehabilitation or for serious, debilitating injuries that require some constant level of care.  The state also has given us information that only 1 or 2 of those deaths were to people without pre-existing illnesses.

What would be really helpful for public understanding and transparency is for both agencies to give us a chart of cases, hospitalizations, deaths by age group and population by age group, all reported as of the same date.  One nice chart with all the data in one place, and then maybe calculate those population fatality rates and the case fatality rates.  Not a hard ask.  And be aware, the per capita, or population based rates can only go up, every new death makes them a little higher.  The case fatality rates will inevitably go down, because with this disease there are so many cases that are symptom-free or mild, that case numbers based on positive infection testing results are far too low.  But one thing that the population data clearly shows, and this pattern is unlikely to change, is the extreme difference between how the epidemic is affecting younger Americans and older ones.