Now that #Indiana is moving beyond vaccinating front-line healthcare workers first (which mostly everyone agreed on) there's a lot of disagreement over who should get the #COVID19 vaccine next. A THREAD [1/x]
Who gets vaccinated next often turns into trying to compare the “value” of one life (a 70yo in a retirement home) vs another (a 40yo teacher in the classroom). In economics we would call this a “normative” question because the answer depends on personal value judgements. [2/x]
By their nature, normative questions DO NOT HAVE a single correct answer. [3/x]
But what if we all agree that the vaccine should be given first to the people that are “most likely to be infected and die from #COVID19.” Who are those people? In economics we call this a “positive” question because there IS a single correct answer now. [4/x]
Take any two people and one MUST have a higher or lower risk (or the same) of dying from #COVID19 than the other. We may disagree what the exact risks are for two different people but can all agree that one person MUST have a higher risk (or they’re equal). [5/x]
Looking at it this way changes it from a normative question to a positive question. We can still argue about the right answer, BUT we all acknowledge that probability doesn’t care and there is a single “correct” answer. [6/x]
Despite this, there are STILL strong disagreements among people over who is “most likely to be infected and die from #COVID19.” Why is that? I think there are two main reasons. [7/x]
FIRST: We're used to thinking about ABSOLUTE risks from #COVID19: “What’s my risk of infection if I go to the grocery store?” or “What’s grandma’s risk of infection if she visits her grandkids?” However, when deciding who to vaccinate, the RELATIVE risk is what’s important. [8/x]
Instead we need to ask “What’s the risk to a 70yo in a locked-down retirement home RELATIVE TO a 50yo elementary school teacher?” Relative risks are harder for most of us to judge because we’re more familiar with our own risks than those of others. [9/x]
SECOND: unless you work with probability theory regularly, calculating the probability of something can be very counter-intuitive. For example, consider a hypothetical discussion that goes something like this: [10/x]
Gene: “Old people are way more likely to die of #COVID19 than most teachers, so we need to vaccinate them first!”

Mike: “Teachers are way more likely to be exposed to #COVID19 so we need to vaccinate them at the same time as old people!” [11/x]
Who's making the “correct” argument here? Kinda both. The probability of someone dying from Covid-19 depends on both the probability of exposure/infection AND the probability of dying once you’re exposed. [12/x]
The probability someone in the general population dies of #COVID19 is given by:

Pr(Dies) = Pr(Infected) x Pr(Dies|Infected)

Where Pr(Infected) is the prob. the person gets infected. Pr(Dies|Infected) is the prob. the person dies GIVEN that they were infected. [13/x]
Gene is saying Pr(Dies|Infected) is huge for old people.
Mike is saying Pr(Infected) is huge for teachers.

Neither is "wrong" but the PRODUCT of these two terms determines the overall probability of someone dying from #COVID19 (given that they’re not currently infected). [14/x]
Also, Pr(Dies|Infected) is easier to measure using observed case fatality rates. Pr(Infected) is way harder to measure. What's the probability of being infected as a teacher? It’s definitely more than someone in a locked-down retirement home, but how much more? [15/x]
To help understand relative risks, here's a table that lets you compare the OVERALL risk of someone dying from Covid-19 in Indiana based on their age and a particular risk of infection. [16/x]
Values in the table are indexed to the overall risk of death RELATIVE TO someone age 70-79 with a 5% risk of infection. So how do you use this table? Here are a few examples of how to interpret the table. [17/x]
Someone aged 60-69 with a 45% risk of infection has a value of 300. This means they’re THREE TIMES more likely (100x3=300) to be infected with Covid-19 and die than someone in the 70-79 age who has a 5% risk of infection. [18/x]
Someone aged 30-39 with a 75% risk of infection has a value of 17. This means they’re 83% less likely (100-17=83) to be infected with Covid-19 and die than someone in the 70-79 age who has a 5% risk of infection. [19/x]
If you aren’t a numbers person, just look at the colors. Yellow reflects ages and situations where the overall risk of death from Covid-19 is similar to that of someone 70-79 with a 5% chance of infection. Red is worse risk of death and green is less risk of death. [20/x]
When you look at it this way, you can see Gene’s point pretty clearly. If you’re a 45yo teacher then even if you’re GUARANTEED to get infected, your overall risk of death is still HALF that of a 75yo locked-down in a retirement home with a very low risk of infection (5%). [21/x]
What’s the bottom line? The relative risk of dying from #COVID19 based on age & risk of infection is complicated and can be counter intuitive. It depends on factors that we all have different experiences with. [22/x]
Someone in the medical field may know first-hand what mortality rates by age are like. Someone in the education field may know first-hand what risk of infection in the classroom is like. [23/x]
CAVEATS: The table is a gross simplification. It ignores sex, ethnicity, high-risk conditions,etc. It looks only at deaths and ignores potential long-term health consequences. It’s not meant to be exhaustive or definitive, but purely illustrative. [END]
If you prefer, you can read a version of this thread as a single page here:
You can follow @MicahPollak.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled:

By continuing to use the site, you are consenting to the use of cookies as explained in our Cookie Policy to improve your experience.